JP7710630B1 - Trajectory generation device and trajectory generation method - Google Patents

Abstract

The technology disclosed in this specification is a technology for suppressing excessive constraints when setting the trajectory of a moving object. A trajectory generation device related to the technology disclosed in this specification includes a constraint condition setting unit for setting constraint conditions that a moving object passing through the trajectory must satisfy, and a trajectory generation unit for generating a trajectory along which the moving object will pass so as to satisfy the constraint conditions. The constraint conditions include those set based on the probability of a state of the moving object that should satisfy a predetermined condition.

Description

The technology disclosed in this specification relates to trajectory generation technology for moving objects.

A method has been proposed for generating a trajectory of a moving object by taking into account the probability of future events. For example, Patent Document 1 shows a method for probabilistically calculating the future behavior of an obstacle and calculating the optimal trajectory of the vehicle.

JP 2020-15489 A

If the constraints imposed when setting a trajectory include events with a low probability of occurring, the opportunities for the vehicle (mobile body) to perform the desired action will be reduced, which may reduce the freedom of action of the mobile body and reduce the comfort of the occupants.

The technology disclosed in this specification has been developed in consideration of the problems described above, and is a technology for suppressing excessive constraints in setting the trajectory of a moving body.

A trajectory generation device that is a first aspect of the technology disclosed in the present specification includes a constraint condition setting unit for setting constraint conditions that must be satisfied by a moving body passing through a trajectory, and a trajectory generation unit for generating the trajectory along which the moving body will pass so as to satisfy the constraint conditions, including a constraint condition that is set based on the probability of a state of the moving body that should satisfy predetermined conditions, and the constraint condition setting unit sets the probability of the state of the moving body under the constraint conditions based on the time difference between the current time and the time at which the moving body will pass through the trajectory generated by the trajectory generation unit .

According to at least the first aspect of the technology disclosed in the present specification, a trajectory is set to satisfy constraints based on the probability of the state of the moving body, thereby making it possible to suppress excessive constraints in setting the trajectory of the moving body.

Furthermore, objects, features, aspects and advantages associated with the technology disclosed in the present specification will become more apparent from the detailed description set forth below and the accompanying drawings.

1 is a diagram conceptually illustrating an example of the configuration of a vehicle control system including a trajectory generation device according to an embodiment. FIG. 13 is a diagram conceptually illustrating a modified example of the configuration of a vehicle control system including a trajectory generation device according to an embodiment. FIG. 13 is a diagram conceptually illustrating a modified example of the configuration of a vehicle control system including a trajectory generation device according to an embodiment. 1 is a diagram illustrating an example of a hardware configuration of a host vehicle equipped with a vehicle control system according to an embodiment. 4 is a flowchart showing an example of a procedure for controlling driving of a host vehicle according to an embodiment; 13 is a flowchart showing an example of a procedure for generating a target trajectory. FIG. 13 is a diagram conceptually illustrating an example of the range of a no-entry area represented by equation (9). FIG. 13 is a diagram illustrating an example of the relationship between a probability to be satisfied and time. FIG. 13 is a diagram showing an example of a probability distribution of no-entry areas in the near future. FIG. 13 is a diagram showing an example of a probability distribution of no-entry areas in the distant future. FIG. 13 is a diagram showing an example of a probability distribution of a modified no-entry area in the distant future. FIG. 13 is a diagram illustrating an example of the relationship between a probability to be satisfied and a distance from a current position. FIG. 13 is a diagram illustrating an example of a relationship between a probability to be satisfied and an action type. 11A and 11B are diagrams showing a no-entry area when an obstacle position is at an expected value and a no-entry area when the obstacle position is at a position with variation; FIG. 13 is a diagram showing a no-entry area after deformation.

Hereinafter, the embodiments will be described with reference to the attached drawings. In the following embodiments, detailed features are shown to explain the technology, but they are merely examples and not all of them are necessarily essential features for the embodiments to be feasible.

Note that the drawings are schematic, and for ease of explanation, components may be omitted or simplified as appropriate in the drawings. Furthermore, the relative sizes and positions of components shown in different drawings are not necessarily described accurately and may be changed as appropriate. Furthermore, hatching may be used in drawings such as plan views that are not cross-sectional views to make it easier to understand the contents of the embodiments.

In addition, in the following description, similar components are illustrated with the same reference numerals, and their names and functions are also the same. Therefore, detailed descriptions of them may be omitted to avoid duplication.

In addition, in the descriptions provided in this specification, when a certain component is described as "comprising," "including," or "having," unless otherwise specified, this is not an exclusive expression that excludes the presence of other components.

In addition, even if ordinal numbers such as "first" or "second" are used in the descriptions in this specification, these terms are used for convenience to facilitate understanding of the contents of the embodiments, and the contents of the embodiments are not limited to the orders that may result from these ordinal numbers.

<Embodiment>
The trajectory generating device and the trajectory generating method according to this embodiment will be described below.

<Configuration of the trajectory generation device>
FIG. 1 is a diagram conceptually showing an example of the configuration of a vehicle control system including a trajectory generation device according to this embodiment.

As shown in the example of FIG. 1, the vehicle control system 500 includes a vehicle control device 100, an information acquisition unit 200, and a controller unit 300.

The vehicle control device 100 comprises a trajectory generating device 110 and a vehicle control unit 120.

The trajectory generation device 110 includes a constraint condition setting unit 113 that sets constraint conditions that a moving body, such as a vehicle, robot, or drone, that travels along the trajectory must satisfy, and a trajectory generation unit 114 that generates a trajectory along which the moving body travels. In this embodiment, a vehicle is assumed as the moving body.

The vehicle control unit 120 controls the operation of the vehicle so that it follows the trajectory generated by the trajectory generation device 110. Specifically, the vehicle control unit 120 outputs a control signal to the controller unit 300 to control the vehicle.

The information acquisition unit 200 acquires information used when the trajectory generation device 110 generates a trajectory, and outputs it to the trajectory generation device 110. The information acquisition unit 200 includes a host vehicle information acquisition unit 210 that acquires information about the host vehicle, an obstacle information acquisition unit 220 that acquires information about obstacles, and a road information acquisition unit 230 that acquires information about roads.

The controller unit 300 controls the operation of the vehicle based on a control signal input from the vehicle control unit 120. The controller unit 300 includes a powertrain controller 310 that controls a powertrain unit described below, a brake controller 320 that controls a brake unit described below, and an EPS (Electric Power Steering) controller 330 that steers the front wheels independently of the driver's operation of the steering wheel 2.

Figure 2 is a conceptual diagram showing a modified configuration of a vehicle control system including a trajectory generation device related to this embodiment.

As shown in the example of FIG. 2, the vehicle control system 500A includes a vehicle control device 100A, an information acquisition unit 200, and a controller unit 300.

The vehicle control device 100A comprises a trajectory generating device 110A and a vehicle control unit 120.

The trajectory generating device 110A includes a constraint condition setting unit 113, a trajectory generating unit 114, and a no-entry area setting unit 112 that sets a no-entry area.

As shown in FIG. 2, the trajectory generating device 110A may additionally be provided with a no-entry area setting unit 112.

Figure 3 is a conceptual diagram showing a modified configuration of a vehicle control system including a trajectory generation device related to this embodiment.

As shown in the example of FIG. 3, the vehicle control system 500B includes a vehicle control device 100B, an information acquisition unit 200, and a controller unit 300.

The vehicle control device 100B includes a trajectory generating device 110B and a vehicle control unit 120.

The trajectory generation device 110B includes a constraint condition setting unit 113, a trajectory generation unit 114, a no-entry area setting unit 112, and an obstacle movement prediction unit 111 that predicts the movement of an obstacle.

The obstacle movement prediction unit 111 predicts the state quantity of an obstacle at a future time. The state quantity of an obstacle includes at least position information, as well as direction, speed, acceleration, yaw rate, etc. Information on the type of behavior, such as turning right or left or changing lanes, may also be included.

As shown in FIG. 3, the trajectory generating device 110B may additionally be provided with a no-entry area setting unit 112 and an obstacle movement prediction unit 111.

Figure 4 is a diagram showing an example of the hardware configuration of a vehicle equipped with the vehicle control system 500 relating to this embodiment.

As shown in the example of Figure 4, the host vehicle 1 includes, as a drive system, a steering wheel 2, a steering shaft 3, a steering unit 4, an EPS motor 5, a powertrain unit 6, and a brake unit 7. The powertrain unit 6 is, for example, a gasoline-fueled engine.

The vehicle 1 also has a sensor system including a forward camera 11, a radar sensor 12, a GNSS (Global Navigation Satellite System) sensor 13, a yaw rate sensor 16, a speed sensor 17, an acceleration sensor 18, a steering angle sensor 20 and a steering torque sensor 21.

Furthermore, the vehicle 1 is equipped with a V2X (Vehicle to Everything) receiver 15, a vehicle control device 100, an EPS controller 330, a powertrain controller 310 and a brake controller 320.

The steering wheel 2, which is installed so that the driver can drive the vehicle, is connected to a steering shaft 3. A steering unit 4 is connected to the steering shaft 3.

The steering unit 4 rotatably supports the two front tires as steering wheels, and is supported on the vehicle body frame so that it can be steered. Therefore, the torque generated by the driver's operation of the steering wheel 2 rotates the steering shaft 3, and the steering unit 4 steers the front wheels to the left and right. By operating the steering wheel 2, the driver can control the lateral movement of the vehicle when moving forward and backward.

The steering shaft 3 can also be rotated by the EPS motor 5. The EPS controller 330 controls the current flowing through the EPS motor 5, allowing the front wheels to be steered independently of the driver's operation of the steering wheel 2.

The vehicle control device 100 is, for example, an integrated circuit such as a microprocessor also known as an ADAS-ECU (Advanced Driving Assistance Systems-Electronic Control Unit), and includes an A/D (Analog/Digital) conversion circuit, a D/A (Digital/Analog) conversion circuit, a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), etc.

The vehicle control device 100 is connected to a forward camera 11, a radar sensor 12, a GNSS sensor 13, a V2X receiver 15, a yaw rate sensor 16 that detects yaw rate, a speed sensor 17 that detects the speed of the host vehicle, an acceleration sensor 18 that detects the acceleration of the host vehicle, a steering angle sensor 20 that detects the steering angle, a steering torque sensor 21 that detects the steering torque, an EPS controller 330, a powertrain controller 310 and a brake controller 320.

The vehicle control device 100 processes information input from various connected sensors according to a program stored in ROM, transmits a target driving force to the powertrain controller 310, and transmits a target braking force to the brake controller 320.

The vehicle control device 100 has the function of calculating the optimal driving route for the destination set by the driver, and stores road information on the driving route. The road information is map node data that represents the road linearity, and can be acquired from the road information acquisition unit 230. Each map node data incorporates information such as latitude, longitude, altitude, lane width, cant angle, and inclination angle that indicate the absolute position at each node.

The forward camera 11 is installed in a position where the dividing line in front of the vehicle can be detected as an image, and detects the environment in front of the vehicle 1, such as lane information and the position of obstacles, based on the image information. Note that in the vehicle control system 500 of this embodiment, only a camera that detects the environment in front of the vehicle 1 is shown, but another camera that detects the environment behind and to the sides of the vehicle 1 may be provided. The forward camera 11 can also be used to estimate the condition of the road surface on which the vehicle 1 is traveling.

The radar sensor 12 outputs the relative distance and relative speed between the vehicle 1 and surrounding vehicles by irradiating a target object with a radar and detecting the reflected wave. The radar sensor 12 can be a well-known distance measuring sensor such as a millimeter wave radar, LiDAR (Light Detection and Ranging), a laser range finder, or an ultrasonic radar.

The GNSS sensor 13 receives radio waves from positioning satellites using an antenna (not shown) mounted on the vehicle, and performs positioning calculations to output the absolute position and absolute orientation of the vehicle.

The V2X receiver 15 has a function of acquiring and outputting information through wireless communication between other vehicles, including surrounding vehicles, and roadside devices and the vehicle 1. The acquired information includes surrounding vehicle information such as the position and speed of the surrounding vehicles relative to the vehicle 1, and road information such as the friction coefficient of the road surface.

The EPS controller 330 controls the driving direction of the vehicle 1 by controlling the EPS motor 5 to achieve the target steering angle transmitted from the vehicle control device 100.

The powertrain controller 310 controls the acceleration of the host vehicle 1 by controlling the powertrain unit 6 so as to achieve the target acceleration force transmitted from the vehicle control device 100.

The brake controller 320 controls the deceleration of the host vehicle 1 by controlling the brake unit 7 to achieve the target braking force transmitted from the vehicle control device 100.

The host vehicle information acquisition unit 210 acquires vehicle information, which is information about the host vehicle 1. The vehicle information includes state quantities of the host vehicle that represent the state of the host vehicle 1. The host vehicle information acquisition unit 210 includes, for example, a GNSS sensor 13, a yaw rate sensor 16, a speed sensor 17, an acceleration sensor 18, a steering angle sensor 20, and a steering torque sensor 21.

The obstacle information acquisition unit 220 acquires obstacle information including position information of surrounding vehicles present around the vehicle 1. The obstacle information acquisition unit 220 is, for example, a forward camera 11, a radar sensor 12, a V2X receiver 15, etc.

The road information acquisition unit 230 acquires road information, which is information about the road on which the vehicle 1 is traveling. The road information acquisition unit 230 is, for example, a forward camera 11 and a V2X receiver 15.

Although a vehicle having only an engine as a driving force source has been shown as the vehicle 1 equipped with the vehicle control system 500 of this embodiment, the vehicle may also have only an electric motor as a driving force source, or may have both an engine and an electric motor as driving force sources.

In this embodiment, the trajectory generation unit 114 generates a trajectory for the moving object by solving an optimization problem. However, the trajectory generation unit 114 may generate a trajectory by other methods, such as a method of generating a trajectory with the center of a road or a route that another vehicle has passed as a target position, or a method of randomly generating multiple candidate routes and selecting one from them using some kind of evaluation index.

<Optimization problem setting>
The trajectory generation unit 114 predicts the vehicle state quantity x from the current time 0 to a prediction period Th into the future at a time interval Ts using a vehicle model f that mathematically represents the motion of the vehicle, and solves an optimization problem to obtain series data of the control input u that minimizes an evaluation function J that expresses a desired operation of the vehicle under constraint conditions.

Then, based on the optimized control input u and vehicle model f obtained from the optimization problem, series data of the optimized vehicle state quantity x is predicted from the current time 0 to the prediction period Th in the future at a time interval Ts.

Then, based on the series data of the optimized control input u and the series data of the vehicle state quantity x, a trajectory ξ, which is series data including the vehicle's position, is generated. In the following description, the time from the current time to the prediction period Th may be abbreviated as the horizon.

<Formulation of the optimization problem>
As described above, in this embodiment, a constrained optimization problem is solved at regular intervals. The optimization problem is formulated as follows.

Here, J is the evaluation function, x is the vehicle state quantity, u is the control input, f is a vector-valued function related to the dynamic vehicle model, and x0 is the initial value (current vehicle state quantity).

Here, the above optimization problem is a constrained optimization problem subject to the constraints described below.

In this embodiment, the above optimization problem is treated as a minimization problem, but it can also be treated as a maximization problem by inverting the sign of the evaluation function.

In this embodiment, the following equation is used for the evaluation function J:

Here, x(k) is the vehicle state quantity at prediction point k (k=0,...,N), and u(k) is the control input at prediction point k (k=0,...,N). h is a vector value function related to the evaluation items, _hN is a vector value function related to the evaluation items at the end (prediction point N), and r(k) is a reference value at prediction point k (k=0,...,N). W and _WN are weighting matrices, which are diagonal matrices having weights for each evaluation item in the diagonal components, and can be changed as appropriate as parameters.

<Vehicle model>
In this embodiment, the vehicle state quantity x and the control input u used in the trajectory generation unit 114 are set as follows.

Here, β is the sideslip angle, γ is the yaw rate, ax is the longitudinal acceleration, δ is the steering angle, axt is the target longitudinal acceleration, and δt is the target steering angle. Additionally, jt is the target longitudinal jerk, and ωt is the target steering angular velocity. Note that as long as the vehicle state quantity x includes a variable related to position, and the variables related to steering and vehicle speed are included in either the vehicle state quantity x or the control input u, the vehicle state quantity x and the control input u may be set in any way. Additionally, the position variable is not limited to the Cartesian coordinate system, and may be defined, for example, in a route coordinate system.

In addition, when the vehicle control device 100 performs only steering control, the vehicle state quantity x includes a variable related to position, and the vehicle state quantity x and the control input u may be set in any manner as long as a variable related to steering is included in either the vehicle state quantity x or the control input u.

The vehicle model f uses the two-wheel model shown below.

Here, M is the vehicle mass and I is the vehicle's yaw moment of inertia. lf and lr are the distances from the front and rear wheel axles to the vehicle's center of gravity. Tax and Tδ are time constants when the tracking ability of the longitudinal acceleration and steering angle to the target values is expressed in a first-order lag system. Yf and Yr are the cornering forces of the front and rear wheels, and are expressed by equations (107) and (108) using the cornering stiffnesses Cf and Cr of the front and rear wheels.

In addition, vehicle models other than the two-wheel model may be used for vehicle model f.

<Operation of the vehicle control device>
FIG. 5 is a flowchart showing an example of a procedure for controlling driving of the host vehicle according to this embodiment.

In step ST110 of FIG. 5, obstacle information is acquired by the obstacle information acquisition unit 220. The obstacle information is information including the positions of obstacles including surrounding vehicles, and in this embodiment, if an obstacle is present to the left front of the host vehicle, the positions of the right front end PFR, right rear end PRR, and left rear end PRL of the obstacle in the host vehicle coordinate system are acquired, and if an obstacle is present to the right front of the host vehicle, the positions of the left front end PFL, left rear end PRL, and right rear end PRR of the obstacle in the host vehicle coordinate system are acquired.

Furthermore, the obstacle information acquisition unit 220 estimates the position of the obstacle's left front end PFL or right front end PFR, the position Xo, Yo of the center PC, the vehicle direction θo, the vehicle speed Vo, the length lo, and the width wo based on this position information.

Next, in step ST120 of FIG. 5, road information is acquired by the road information acquisition unit 230. The road information is information including the boundaries of the road on which the vehicle is traveling and the adjacent roads (hereinafter referred to as the vehicle's lane, left lane, and right lane), and in this embodiment, the coefficients when the left and right dividing lines of the vehicle's lane, left lane, and right lane are expressed as a third-order polynomial are acquired. That is, for the dividing line to the left of the vehicle's lane (which is also the right dividing line of the left lane), the values of cel0 to cel3 of the following equation are acquired.

For the right dividing line of the own lane (which is also the left dividing line of the right lane), the values of cer0 to cer3 are obtained using the following formula.

For the left dividing line of the left lane, the values cl10 to cl13 are obtained using the following formula.

For the right dividing line of the right lane, the values of crr0 to crr3 are obtained using the following formula.

In this case, the center of the own lane, the center of the left lane, and the center of the right lane are expressed by equations (205), (206), and (207), respectively.

Here, each coefficient is expressed by equation (208), equation (209), and equation (210).

In addition, the information on the dividing lines is not limited to third-order polynomials and may be expressed using any function.

In step ST130 of FIG. 5, vehicle information is acquired by the host vehicle information acquisition unit 210. The vehicle information is information such as the steering angle, yaw rate, speed, and acceleration of the host vehicle, and in this embodiment, the steering angle δ, yaw rate γ, speed V, and longitudinal acceleration ax are acquired.

Next, in step ST210 of Fig. 5, obstacle movement prediction unit 111 (Fig. 3) predicts the movement of the obstacle. In the movement prediction, the center position _Xo (k), _Yo (k), vehicle direction _θo (k), and vehicle speed _Vo (k) of the obstacle at each prediction point k (k = 0, ..., N) are predicted. In this embodiment, the movement of the obstacle is approximated by uniform linear movement, and the center position _Xo (k), _Yo (k), vehicle direction _θo (k), and vehicle speed _Vo (k) of the obstacle at the prediction point k (k = 0, ..., N) are predicted as follows:

where _Xo (0), _Yo (0), _θo (0), and _Vo (0) are the center position, vehicle direction, and vehicle speed of the obstacle at the current time acquired by the obstacle information acquisition unit 220. When there are multiple obstacles, the above prediction is performed for each obstacle. Note that instead of uniform linear motion, a prediction may be made that the obstacle moves at a uniform speed along the driving lane. Alternatively, a driver model may be used to make the prediction.

On the other hand, the obstacle movement prediction unit 111 (Figure 3) may predict the future behavior of the obstacle and display it probabilistically. By displaying the behavior of the obstacle probabilistically, the obstacle movement prediction unit 111 can predict the behavior of the obstacle including the variability of the behavior. Note that the obstacle movement prediction unit 111 may display only a part of the state quantities of the obstacle probabilistically. Methods for expressing the behavior of an obstacle probabilistically include a probability distribution according to a Gaussian distribution, a probability distribution according to a uniform distribution, and a probability distribution up to predetermined upper and lower limits.

Next, in step ST220 of FIG. 5, the no-entry area setting unit 112 (FIGS. 2 and 3) sets a no-entry area ζ. The no-entry area ζ may be outside the lane of the road, or based on a wall or an obstacle.

In this embodiment, an elliptical no-entry area is set at the center positions _Xo (0) and _Yo (0) of the obstacle at each prediction point k (k=0, ..., N). The equation of the ellipse ζ(X, Y)=0 is expressed by the following equation.

l _a and l _b are the lengths of the major and minor axes of an ellipse set for an obstacle, and may be changed for each prediction point k. The center of the ellipse does not need to coincide with the central positions X _o (k) and Y _o (k) of the obstacle. The no-entry area set for an obstacle does not need to be elliptical, and any shape of no-entry area may be set. When there are multiple obstacles, a no-entry area is set for each obstacle.

The forbidden area ζ may be one in which the behavior of obstacles forming the area is represented probabilistically, or the position or size of the area may change with time. In this case, the forbidden area _ζt may be represented, for example, as in Equation (1) described below.

When considering safety, it is more important to reduce false positives (mistakenly determining that a lane change is possible when it is not) than false negatives (mistakenly determining that a lane change is possible when it is not) when determining whether a lane change is possible. Therefore, if you want to reduce false positives when determining whether a lane change is possible, you can expand the no-entry area only when making a decision. This makes it more difficult for the vehicle to reach the target lane at the time of decision, so if there is not enough time to change lanes, it can be determined that a lane change is not possible, improving safety.

Next, in step ST230 of FIG. 5, the trajectory generation unit 114 generates a target trajectory ξ by solving the optimization problem of equation (101). The target trajectory ξ is series data including the target position of the host vehicle, and in this embodiment, it is the series data of the vehicle state quantity x of equation (104).

The trajectory generation unit 114 generates a target trajectory ξ for lane keeping (target lane keeping trajectory ξLK) when the target action is lane keeping, and generates a target trajectory ξ for lane change (target lane change trajectory ξLC) when the target action is lane change.

Next, in step ST240 of FIG. 5, the trajectory generation unit 114 determines whether or not to change lanes based on the target lane change trajectory ξLC. In this embodiment, this determination is made when the target behavior changes to a lane change, that is, when the lane change starts. However, the determination may also be made while the lane change is continuing.

If it is determined that a lane change is possible, the target lane change trajectory ξLC is output to the vehicle control unit 120. On the other hand, if it is determined that a lane change is not possible, the target lane keeping trajectory ξLK is output to the vehicle control unit 120.

In this embodiment, the target lane-keeping trajectory ξLK output at this time is generated based on the target lane-keeping trajectory ξLK last output by the trajectory generation unit 114. Note that it is also possible to output the target lane-keeping trajectory ξLK obtained by changing the target behavior to lane keeping and resolving the optimization problem. Furthermore, if it is determined that a lane change is not possible and there is sufficient calculation time, the target lane-changing trajectory ξLC may be recalculated by changing the reference value or weight of the optimization problem.

If the target action is lane keeping, no judgment is made and the target lane keeping trajectory ξLK generated by the trajectory generation unit 114 is output as is.

Next, in step ST250 of Fig. 5, the vehicle control unit 120 calculates target values for steering control and vehicle speed control so that the vehicle follows the target trajectory ξ. In this embodiment, the vehicle control unit 120 calculates a target steering angle δt, which is a target value related to steering, and a target longitudinal acceleration axt, which is a target value related to vehicle speed. In this embodiment, since the target trajectory ξ includes the optimal value of the target steering angle _δt (k) and the optimal value of the target longitudinal acceleration _axt (k) at each prediction point k (k = 0, ..., N), the target steering angle _δt and the target longitudinal acceleration axt are calculated by interpolating the optimal value of the target steering angle _δt ( _k ) and the optimal value of the target longitudinal acceleration axt(k) in the time direction according to the control cycle of each actuator.

5, the powertrain controller 310, the brake controller 320, and the EPS controller 330 control the actuators based on the control amount. In this embodiment, the EPS motor 5 is controlled so that the steering angle δ follows the target steering angle δt, and the powertrain unit 6 and the brake unit 7 are controlled so that the longitudinal acceleration ax follows the target longitudinal acceleration axt.

<Procedure for generating target trajectory>
6 is a flowchart showing an example of a procedure for generating a target trajectory, which is performed in step ST230 in FIG.

First, in step ST231 of Fig. 6, the trajectory generating unit 114 calculates a group of reference points. Here, the group of reference points is series data of reference positions _Xr , _Yr , reference trajectory direction _ψr , and reference vehicle speed _Vr from the current time 0 to a prediction period _Th in the future with a time interval _Ts . Hereinafter, the series data of reference positions _Xr (k), _Yr (k) (k = 0, ..., N) is referred to as a reference trajectory χr.

The reference positions _Xr (k), _Yr (k), reference track orientation _ψr (k), and reference vehicle speed _Vr (k) (k=0, . . . , N) at each time are determined as follows.

First, the reference vehicle speed _Vr (k) is determined based on the speed limit Vl of the lane and the vehicle speed Vp of the preceding vehicle, for example, _Vr (k)=Vl. Note that _Vr (k) does not need to be a constant value within the horizon.

Next, when the target behavior is lane keeping, the reference positions _Xr (k), _Yr (k) and the reference trajectory direction _ψr (k) are determined based on the X position, Y position and trajectory direction of the lane center so that the host vehicle can travel in the center of the target lane. At the same time, conditions are set for the relationship between the reference positions _Xr (k), _Yr (k) and the reference vehicle speed _Vr (k) so that the reference positions _Xr (k), _Yr (k) and the reference vehicle speed _Vr (k) are consistent. In other words, the reference positions _Xr (k), _Yr (k) are determined so as to satisfy the following two equations:

Equation (301) is the condition for the reference positions _Xr (k), _Yr (k) to exist on the function Y=l _e (X) (equation (205)) representing the center of the own lane, and equation (302) is the condition for the distance between adjacent reference positions _Xr (k-1), _Yr (k-1) and _Xr (k), _Yr (k) to be equal to the amount of movement of the own vehicle during time interval Ts.

By calculating the direction of the own lane center Y=l _e (X) at the reference positions X _r (k) and Y _r (k) determined by these, the reference trajectory direction ψ _r (k) can also be determined. In addition, hereinafter, the reference trajectory for lane keeping will be referred to as the reference lane keeping trajectory χrLK.

When the target behavior is a lane change, for example, a function Y=l LC (X) is generated that represents a reference trajectory for lane change (reference lane change trajectory χrLC) by connecting the center of the current lane to the center of the target _lane in a continuous and smooth manner.

This reference lane change trajectory χrLC is a trajectory generated without the constraint of not entering a no-entry area, and can be said to be a lane change trajectory when there are no obstacles. A known method such as a spline curve or a quintic function is used for the connection. Then, the following equation is used instead of equation (301) to determine the reference positions _Xr (k) and _Yr (k).

By calculating the orientation of the reference lane change trajectory Y=l _LC (X) at the reference positions _Xr (k) and _Yr (k) determined by these, the reference trajectory orientation _ψr (k) can also be determined. Note that, when connecting, in order to generate a reference lane change trajectory χrLC that completes lane change within the target required time tLC for lane change, for example, the connection is made so that the lateral movement to the target lane is completed within a distance d that the vehicle moves vertically during the target required time tLC. The distance d may be calculated by integrating the reference vehicle speed Vr, or may be calculated by multiplying the current vehicle speed V0 and the target required time tLC. In addition, when the driving lane is curved, the connection may be made in a route coordinate system.

In addition, when there is no need to specify the target required time tLC for a lane change and the prediction period Th is sufficiently long, the reference lane change trajectory χrLC may not be generated, and the reference positions _Xr (k), _Yr (k) may simply be determined using the following equation instead of equation (301):

Here, Y=l _o (X) is a function expressing the center of the target lane, and from equations (205), (206), and (207), l _o =le, ll, and lr when the target lane is the own lane, the left lane, and the right lane, respectively.

The reference positions _Xr (k), _Yr (k), reference track orientation _ψr (k), and reference vehicle speed _Vr (k) (k=0, . . . , N) calculated as above are defined as a group of reference points.

Next, in step ST232 of Fig. 6, the constraint condition setting unit 113 sets constraint conditions. The constraint conditions include, at least in part, probability constraints that include an element based on the probability p _t of a vehicle state that should satisfy a predetermined condition. Note that the constraint conditions may include a constraint condition that does not include a probability (a constraint condition that is not a probability constraint) at the same time as the probability constraint.

The probability constraint is set based on the probability p _t of the vehicle state satisfying a predetermined condition at a future time, or based on the probability p _t of the vehicle state satisfying a predetermined condition over the entire time series. Note that even if the probability constraint is set based on the probability of not satisfying the predetermined condition, it is essentially the same as the above.

A probability constraint may describe the probability of a vehicle state that should satisfy predetermined conditions directly as a constraint, or it may describe the probability as a constraint by replacing the probability with an equivalent equation or an approximation equation.

Examples of probability constraints include, for example, "the probability p _t that the vehicle is located outside the no-entry area ζ is within a threshold value" or "the probability p _t that the state quantity of the vehicle is within a predetermined range is within a threshold value."

If the probability constraint is "at each time, the probability that the acceleration of the vehicle is within a predetermined range is greater than a threshold value," then if the acceleration of the vehicle at time t is a _t , the lower limit of the acceleration is a _min , the upper limit of the acceleration is a _max , the probability of event A occurring is P[A], and the probability that the condition should be satisfied is p _t , then the probability constraint can be expressed as follows:

Probability constraints based on the state quantities of the vehicle may be set based on at least one of the upper and lower limits of the state quantities of the vehicle (steering angle, steering angle velocity, lateral acceleration, lateral deviation from the center of the trajectory, etc.). Even when the vehicle travels along the generated trajectory, there may be deviations from the trajectory due to disturbances (wind, road resistance, slopes), modeling errors, etc., so the future state quantities of the vehicle will vary and change probabilistically.

When the probability constraint is "the probability that the host vehicle is located outside the no-entry area ζ at each time is greater than a threshold value", the position of the host vehicle at time t is ( _xt , _yt ), the predicted position of the obstacle (vehicle) is ( _Xobst , _Yobst ), the no-entry area corresponding to the predicted position ( _Xobst , _Yobst ) of the obstacle (vehicle) is the no-entry area _ζt , the probability of an event A occurring is P[A], and the probability that a condition should be satisfied at time t is _pt , the probability constraint can be expressed as follows:

7 is a conceptual diagram showing an example of the range of the no-entry area _ζt shown in equation (9). In FIG. 7, the no-entry area _ζt corresponding to the predicted position of an obstacle is shown, and the position ( _xt , _yt ) of the host vehicle at a future time is located outside the no-entry area _ζt .

In this embodiment, the probability that the center of gravity positions _Xg (k), Yg( _k ) of the vehicle at each predicted point k (k=0,...,N) will not enter the no-entry area _ζ (no-entry area ζt) set in step ST220 (in other words, the probability that the vehicle will be located outside the no-entry area) is set as a constraint condition (above formula (9)). Methods for expressing the center of gravity position of the vehicle probabilistically include a probability distribution according to a Gaussian distribution centered on each predicted point k, a probability distribution according to a uniform distribution, and a probability distribution up to predetermined upper and lower limits.

In this embodiment, the constraint conditions are set based on the no-entry area, but the constraint conditions may be set without using the no-entry area (Figure 1). One example is a probability constraint based on the state quantity of the host vehicle, as explained using equation (8).

Next, in step ST233 of Fig. 6, the trajectory generating unit 114 sets the evaluation function J (equation (103)). In this embodiment, the target trajectory ξ for the host vehicle to follow the reference point group (reference position _Xr (k), _Yr (k), reference trajectory orientation _ψr (k), reference vehicle speed _Vr (k) (k = 0,...,N)) calculated in step ST231 can be generated, and vector value functions h and hN related to the evaluation items are set as follows so that the control input at that time is small.

_ew (k) is the lateral deviation with respect to the reference positions _Xr (k), _Yr (k) at the prediction point k (k = 0, ..., N), and is expressed as shown in equation (309) using the reference positions _Xr (k), _Yr (k) at the prediction point k (k = 0, ..., N) and the reference orbit orientation _ψr (k).

Furthermore, the reference values r(k) and r(N) are set as follows:

Here, _Vr (k) is the reference vehicle speed. This allows the trajectory generation unit 114 to generate a target trajectory that allows the host vehicle to follow the reference point group with a small control input. In order to improve the followability to the reference point group and the ride comfort, the trajectory direction, yaw rate, longitudinal acceleration, lateral acceleration, etc. may be added to the evaluation items. In addition, the evaluation function may be changed depending on the target behavior.

Next, in step ST234 of FIG. 6, the trajectory generation unit 114 calculates the optimal control input u* by solving the constrained optimization problem equation (101) using the evaluation function equation (103) and the constraint condition equation (9). The optimal control input u ^* is calculated using a known means such as ACADO (Automatic Control and Dynamic Optimization) developed by K. U. Leuven University and AutoGen, which is an automatic ^code generation tool that solves optimization problems based on the C/GMRES method. When ACADO or AutoGen is used, a time series (optimal control input) u ^* of the optimized control input at each prediction point k (k=0, . . . , N-1) is output. That is, the output of step ST234 is equation (312).

Here, _jxt ^* (k) and _ωt ^* (k) (k=0,...,N-1) are optimal values of the target pitch jerk and the target steering angular velocity. Note that the solution may be a value that makes the evaluation function fall below a predetermined threshold value, or, if the evaluation function does not fall below the threshold value within a predetermined number of iterations, a value that minimizes the evaluation function within the number of iterations may be the solution.

Next, in step ST235 of Fig. 6, the trajectory generating unit 114 calculates the optimal state quantity x ^* . In the calculation of the optimal state quantity x ^* , the optimal control input u ^* and the vehicle model f are used to calculate a time series (optimum state quantity) x ^* of the optimized vehicle state quantity at each prediction point k (k = 0, ..., N). Therefore, the output of step ST235 is given by equation (313).

Here, _Xg ^* (k), _Yg ^* (k), θ ^* (k), β ^* (k), γ ^* (k), V ^* (k), ax*(k), _axt ^* (k) _, δ ^* (k), and ^{δt *} (k) are the optimum value of the center of gravity position, the optimum value of the vehicle body orientation, the optimum value of the sideslip angle, the optimum value _of the yaw rate, the optimum value of the vehicle speed, the optimum value of the longitudinal acceleration, the optimum value of the target longitudinal acceleration, the optimum value of the steering angle, and the optimum value of the target steering angle, respectively.

Next, in step ST236 of FIG. 6, the trajectory generating unit 114 generates a target trajectory ξ. The target trajectory ξ is generated based on the optimal state quantity x ^* and the optimal control input u ^* . When the trajectory generating unit 114 judges whether or not a lane change is possible based on the information on the position of the target trajectory ξ, the target trajectory ξ only needs to include the optimal center of gravity positions _Xg ^* and _Yg ^* , and when the trajectory generating unit 114 further judges whether or not a lane change is possible based on the steering behavior, the target trajectory ξ only needs to further include the optimal steering angle δ ^* and the optimal target steering angular velocity _ωt ^* , which are variables related to steering. In this embodiment, the optimal state quantity x ^* is set as the target trajectory ξ. Therefore, the output of step ST236 is given by equation (314).

In addition, the target trajectory ξ when the target action is to keep the lane is called the target lane keeping trajectory ξLK, and the target trajectory ξ when the target action is to change the lane is called the target lane change trajectory ξLC.

As described in step ST231, when the target behavior is different, at least the reference trajectory χr is different. However, in addition to that, the item or value of the constraint may be changed in step ST232, or the item or value of the evaluation function may be changed in step ST233.

<How to set the probability p _t >
As a method for setting the probability p _t that should satisfy the conditions in the probability constraint, for example, the following can be considered.

First, a setting based on time is considered. When a probability constraint is set so that a predetermined condition is satisfied at each time, the magnitude of the probability p _t of satisfying the condition may be changed based on a future time. For example, the further in the future the time is, the lower the probability of satisfying the condition may be.

Fig. 8 is a diagram showing an example of the relationship between the probability p _t to be satisfied and time. In Fig. 8, the vertical axis shows the probability p _t to be satisfied, and the horizontal axis shows time. As shown in the example in Fig. 8, the probability p _t to be satisfied can be set lower as the time becomes further in the future from the current time.

Figure 9 shows an example of the probability distribution of no-entry areas in the near future. In Figure 9, the vertical axis shows the probability density, and the horizontal axis shows the position. The area shown by diagonal lines in Figure 9 corresponds to the no-entry areas.

In contrast, Fig. 10 is a diagram showing an example of the probability distribution of forbidden areas in the distant future. In Fig. 10, the vertical axis indicates the probability density, and the horizontal axis indicates the position. The range indicated by diagonal lines in Fig. 10 corresponds to the forbidden areas, and the probability density is lower and the position range is wider than in Fig. 9. In other words, the variation of the forbidden areas is larger. Therefore, if the probability p _t to be satisfied is set under the same conditions as in Fig. 9, the probability constraint will be imposed including forbidden areas with a sufficiently low probability density, and the range in which the probability constraint is satisfied (the white area in Fig. 10) will be excessively narrow.

On the other hand, FIG. 11 is a diagram showing an example of the probability distribution of the changed no-entry area in the distant future. In FIG. 11, the vertical axis indicates the probability density, and the horizontal axis indicates the position. The range indicated by the diagonal lines in FIG. 11 corresponds to the changed no-entry area, and the spread of the probability density is the same as in FIG. 10, but since the range is changed so that the range with a low probability density is not included in the no-entry area, the position range is about the same as in FIG. 9. The change corresponds to setting the probability p _t to be satisfied low in the distant future. This prevents the variation in the no-entry area from becoming excessively large, and prevents the range (the white area in FIG. 11) that satisfies the probability constraint from becoming excessively narrow. As a result, the chances of executing the target action of the host vehicle are increased.

Next, a setting based on distance can be considered. The probability of the state of the vehicle at a position on the trajectory generated by the trajectory generation unit 114 may be set based on the distance between the position on the trajectory and the current position of the vehicle.

Fig. 12 is a diagram showing an example of the relationship between the probability p _t to be satisfied and the distance from the current position. In Fig. 12, the vertical axis shows the probability p _t to be satisfied, and the horizontal axis shows the distance from the current position. As shown in the example in Fig. 12, the probability p _t to be satisfied can be set lower as the distance of the position on the track from the current position of the vehicle increases.

Since the probability density of the trajectory position varies more as the distance from the current position increases, the probability p _t to be satisfied is set low to prevent the range in which the probability constraint is satisfied from being excessively narrowed by events with low probability density, thereby increasing the chances of executing the target action of the host vehicle.

Next, a probability p _t that a condition is satisfied may be set based on at least one of the type of action currently being performed by the host vehicle and the type of action that the host vehicle is scheduled to perform in the future.

Here, in the case of a vehicle, the types of actions include maintaining the current state, following the lane, changing lanes, turning right or left, stopping, avoiding obstacles, parking, etc. In addition, the actions currently being performed by the vehicle and the actions it plans to perform in the future can be determined by finite state machines, ontology, decision trees, reinforcement learning, Markov decision processes, etc.

According to the above, for example, in the case of an action in which approaching an obstacle is undesirable, the probability of approaching the obstacle can be changed, thereby improving comfort while driving.

Furthermore, when the vehicle's current behavior is different from the behavior it plans to perform in the future, the probability p _t to be satisfied may be set high in consideration of the fact that fluctuations increase with the change in behavior. For example, when the current behavior type is lane following and the next behavior type planned to be performed is lane changing, the probability p _t to be satisfied may be set high.

Fig. 13 is a diagram showing an example of the relationship between the probability p _t to be satisfied and the type of behavior. In Fig. 13, the vertical axis indicates the probability p _t to be satisfied, and the horizontal axis indicates time. As shown in the example in Fig. 13, the probability p _t to be satisfied is set lower as the time becomes further in the future, but when the type of target behavior (behavior planned to be performed in the future) is different from the type of behavior currently being performed, the probability p _t to be satisfied is set higher than when the types are the same.

According to the above, when the behavior type is different, the constraint conditions become stricter, and since the change in behavior type is made under such strict constraint conditions, the probability of success (probability of achieving safe driving) is increased, thereby improving the comfort of the drive.

Note that, from the above, the trajectory generation unit 114 can determine that the next action type is executable when a trajectory that satisfies the constraint conditions is obtained. On the other hand, when a trajectory that satisfies the constraint conditions is not obtained, the trajectory generation unit 114 can determine that the next action type is not executable.

Next, a setting based on the degree of necessity is considered. The probability p _t to satisfy the condition may be set based on the degree of necessity of the behavior type. For example, the higher the degree of necessity of the behavior type, the lower the probability p _t to be satisfied may be set. Here, the degree of necessity can be calculated, for example, based on the relationship between the road structure and the behavior type of the vehicle. For example, when traveling on a merging road, the degree of necessity of lane change is increased as the vehicle approaches the merging end. Also, when there is a branching road on the track and the vehicle is traveling on a lane that is not a branching lane, the degree of necessity of lane change is increased as the vehicle approaches the branching point. Also, when a vehicle wants to turn right at an intersection, but the required time is not significantly different if the vehicle turns right at the next intersection, the degree of necessity of turning right is reduced.

<Variations of probability constraints>
The conditional expression of the probability constraint may be modified, for example, by converting it into an equivalent expression, an approximation expression, or a stricter conditional expression.

For example, the probability constraint corresponding to "at each time, the probability that the acceleration of the host vehicle is within a predetermined range is greater than a threshold value" is shown below.

On the other hand, if the condition can be transformed into a speed condition, the vehicle speed can be _Vt , the lower limit of the speed can be Vp _,min , and the upper limit of the speed can be Vp _,max , and the condition can be transformed as follows:

Furthermore, the probability constraint corresponding to "at each time, the probability that the vehicle is located outside the no-entry area is greater than a threshold value" is assumed to be shown below.

On the other hand, taking into consideration the variance in the predicted positions of the obstacles, the conditions can be rewritten as a post-transformation no-entry area, which is an area in which the host vehicle should not enter in order to satisfy the above formula. If this post-transformation no-entry area is ζ _P,t , the probability constraint is expressed as follows in a form that does not include probability.

Specifically, the following transformation is performed: First, the obstacle forbidden area ζ _t is set to the inside of the ellipse shown below.

By capturing an obstacle probabilistically, the predicted position ( _Xobst , _Yobst ) of the obstacle varies, and the expected value of the obstacle position is obtained as a Gaussian distribution with (μ _Xobst , μ _Yobst ) and standard deviation (σ _Xobst , σ _Yobst ).

Fig. 14 is a diagram showing the no-entry area _ζt when the obstacle position is at the expected value (μ _Xobst , μ _Yobst ) and the no-entry area _ζ1t when the obstacle position is at a position ( _X1obst , _Y1obst ) with variation. As shown in Fig. 14, the no-entry area _ζt and the no-entry area _ζ1t have different ranges depending on the variation in the obstacle position. In this case, _dx and _dy, which determine the size of the no-entry area ζt _, are treated as deterministic.

To transform the equation in this case, first, a region where the obstacle position exists with a probability _pt is found. If the expected value (μ _Xobst , μ _Yobst ) = (0, 0) and the standard deviation (σ _Xobst , σ _Yobst ) = (1, 1) according to the characteristics of the Gaussian distribution, this region is inside a circle of radius r centered at the origin, which is found by the following equation. If the expected value is (μ _Xobst , μ _Yobst ) and the standard deviation is (σ _Xobst , σ _Yobst ), the region is inside an ellipse centered at (μ _Xobst , μ _Yobst ) and multiplied by σ _Xobst in the x direction and σ _Yobst in the y direction.

The equation for the two-dimensional standard Gaussian distribution f(x, y) is written as follows:

On the other hand, as a general property of the Gaussian distribution, the cumulative distribution function P(r) is expressed as follows:

When the standard deviation is (1, 1), the range that exists with probability p _t is obtained by solving P(r)=p _t as follows.

Therefore, when the expected value of the predicted obstacle position ( _Xobst , _Yobst ) is obtained as a Gaussian distribution with ( _μXobst , _μYobst ) and standard deviation ( _σXobst , _σYobst ), the range in which the obstacle position exists with probability _pt is inside the ellipse expressed by the following formula.

When the no-entry area for an obstacle is expressed by equation (1) and the obstacle position ( _Xobst , _Yobst ), which is its center, moves inside the ellipse expressed by equation (6), the range of the area that the no-entry area can take can be approximately expressed as the inside of the following ellipse.

Therefore, the transformed no-entry area ζ P, _t, which is an area into which the vehicle position must not enter because the probability that the vehicle position is outside the no-entry area ζ _t of the obstacle is greater than p _t , is inside the ellipse expressed by equation (7).

15 is a diagram showing the post-deformation forbidden area ζ _P,t , As shown in Fig. 15, the post-deformation forbidden area ζ _P,t extends to the outside of the forbidden area ζ _t .

By transforming the condition equations as above (either equation transformation or approximation transformation), the probability constraint equation becomes an equation that is easy for computers to handle, expressed in terms that do not include probability, while maintaining the effect of imposing probability constraints by probabilistically viewing the moving body (own vehicle). This reduces the computational load and improves the computation speed.

In this embodiment, a target trajectory ξ is generated based on the acquired vehicle information, and vehicle control is performed after determining whether or not to change lanes based on the target trajectory ξ. However, the vehicle state quantities may be estimated in advance using known techniques such as a low-pass filter, an observer, a Kalman filter, or a particle filter, and a decision may be made regarding the target action to be taken by the vehicle and the target lane in which the vehicle should travel based on obstacle information, road information, and vehicle information (vehicle state quantities).

For the above decision-making, known techniques such as finite state machines, ontology, decision trees, reinforcement learning, and Markov decision processes can be used. In this embodiment, a finite state machine is used for decision-making, and when autonomous driving begins, the target behavior is to maintain the lane, and the need for a lane change is determined based on the destination and the current lane in which the vehicle is traveling, and the target behavior can be a lane change. In addition, the need for overtaking the vehicle may be determined based on movement prediction information, and the target behavior may be a lane change if overtaking is necessary. In addition, when the target behavior is a lane change, it is also determined whether to change lanes to the right or to the left. This determination can be made based on, for example, the position of the overtaking lane.

For example, if the target action is to stay in lane, the vehicle's own lane is the target lane. If the target action is to change lanes to the right, the right lane is the target lane. However, the moment the vehicle crosses the dividing line and moves into the right lane while changing lanes, the target lane becomes the right lane as seen from the original lane, i.e., the vehicle's own lane after crossing the dividing line. The same applies for changing lanes to the left.

<Effects of the above-described embodiment>
Next, examples of effects produced by the above-described embodiments are shown. In the following description, the effects are described based on the specific configurations shown as examples in the above-described embodiments, but they may be replaced with other specific configurations shown as examples in the present specification as long as the same effects are produced. In other words, for convenience, only one of the corresponding specific configurations may be described as a representative below, but the representatively described specific configuration may be replaced with another corresponding specific configuration.

According to the embodiment described above, the trajectory generation device includes a constraint condition setting unit 113 for setting constraint conditions that a moving body (host vehicle 1) passing through the trajectory must satisfy, and a trajectory generation unit 114 for generating a trajectory along which the host vehicle 1 passes so as to satisfy the constraint conditions. The constraint conditions are set based on the probability of the state of the host vehicle 1 satisfying the predetermined conditions.

With this configuration, a trajectory is set to satisfy constraint conditions based on the probability of the state of the moving body (vehicle), and since the moving body is perceived with a probabilistic spread, it becomes easier to satisfy the constraint conditions, and it becomes possible to set a trajectory that does not impose excessive constraints on unlikely events, thereby increasing the opportunities for the vehicle to perform the desired action.

Furthermore, the same effect can be achieved even if other configurations, examples of which are shown in this specification, are added to the above configuration as appropriate, i.e., other configurations in this specification that were not mentioned as the above configuration are added as appropriate.

Furthermore, according to the embodiment described above, the trajectory generation device includes an entry-prohibited area setting unit 112 for setting an entry-prohibited area ζ (or entry-prohibited area ζ _t ) which is an area where the entry of the host vehicle 1 is prohibited. The constraint condition is set based on the probability that the host vehicle 1 is outside the entry-prohibited area. According to such a configuration, whether or not the constraint condition is satisfied is determined based on the probability that the host vehicle is located outside the entry-prohibited area, so that it is possible to set a trajectory that does not impose excessive constraints on unlikely events, and it is possible to increase the opportunities for the host vehicle to perform the intended action.

Moreover, according to the embodiment described above, the trajectory generation device includes an obstacle movement prediction unit 111 for probabilistically predicting the state of an obstacle. Then, the no-entry area setting unit 112 sets the no-entry area ζ _t based on the state of the obstacle probabilistically predicted by the obstacle movement prediction unit 111. According to such a configuration, the no-entry area is set based on the probabilistically indicated position of the obstacle, and as a result, the no-entry area is captured probabilistically, making it possible to set a trajectory that is not excessively restricted by unlikely events, and thus increasing the opportunities for the host vehicle to perform a desired action.

In addition, according to the embodiment described above, the constraint condition setting unit 113 transforms a term indicating the probability of the state of the vehicle 1 in a condition equation indicating a constraint condition into a term that does not include a probability. With such a configuration, the calculation load is reduced and the calculation speed is improved. Also, a trajectory can be generated using a conventional deterministic calculation method.

Furthermore, according to the embodiment described above, the constraint condition setting unit 113 sets the probability of the state of the host vehicle 1 under the constraint condition based on the time difference between the current time and the time when the host vehicle 1 passes through the trajectory generated by the trajectory generation unit 114. With such a configuration, it is possible to prevent the range that satisfies the probability constraint from becoming excessively narrow. As a result, since there is no excessive guarantee in the distant future, the opportunities for the host vehicle to execute the target action are increased.

Furthermore, according to the embodiment described above, the constraint condition setting unit 113 sets the probability of the state of the host vehicle 1 under the constraint condition lower the greater the time difference between the current time and the time when the host vehicle 1 passes through the trajectory generated by the trajectory generation unit 114. With such a configuration, it is possible to prevent the range that satisfies the probability constraint from becoming excessively narrow. As a result, there is no excessive guarantee in the distant future, and the opportunities for the host vehicle to execute the target action increase.

Furthermore, according to the embodiment described above, the constraint condition setting unit 113 sets, in the constraint condition, the probability of the state of the vehicle 1 at a position on the trajectory generated by the trajectory generation unit 114 based on the distance between the position on the trajectory and the current position of the vehicle 1. With this configuration, events with low probability in the distance are not over-guaranteed, and the chances of the vehicle being able to execute the target action are increased.

Furthermore, according to the embodiment described above, the constraint condition setting unit 113 sets the probability of the state of the vehicle 1 at the position on the trajectory to be lower the greater the distance between the position on the trajectory and the current position of the vehicle 1. With this configuration, there is no over-guarantee for events with low probability in the distance, and the chances of the vehicle being able to execute the target action are increased.

Furthermore, according to the embodiment described above, the constraint condition setting unit 113 sets the probability of the state of the vehicle 1 under the constraint condition based on at least one of the current action type of the vehicle 1 and the next action type of the vehicle 1. With this configuration, for example, in an action in which approaching an obstacle is undesirable, the probability of approaching the obstacle can be changed, improving comfort.

Furthermore, according to the embodiment described above, the constraint condition setting unit 113 sets the probability of the state of the vehicle 1 under the constraint conditions higher when the current action type of the vehicle 1 and the next action type to be performed by the vehicle 1 are different than when they are the same. With this configuration, the constraint conditions become stricter when the action types are different, and since the change in action type is made under such strict constraint conditions, the probability of success (probability of realizing safe driving) is increased, thereby improving driving comfort.

In addition, according to the embodiment described above, the constraint condition setting unit 113 sets the probability of the state of the vehicle 1 under the constraint condition based on the degree of necessity of the action type. With this configuration, an action with a high degree of necessity can be executed even if the probability of success is low.

Furthermore, according to the embodiment described above, the constraint condition setting unit 113 sets the degree of necessity of the behavior type based on the relationship between the trajectory structure at the position on the trajectory generated by the trajectory generation unit 114 and the behavior type of the vehicle 1. With this configuration, it is possible to prioritize the realization of behaviors with a high degree of necessity.

Furthermore, according to the embodiment described above, if the trajectory generation unit 114 obtains a trajectory that satisfies the constraint conditions, it determines that the next type of behavior to be performed by the vehicle 1 is executable. Furthermore, if the trajectory generation unit 114 does not obtain a trajectory that satisfies the constraint conditions, it determines that the next type of behavior to be performed by the vehicle 1 is not executable. With this configuration, it is possible to determine the feasibility of the next type of behavior based on whether or not a trajectory that satisfies the constraint conditions has been obtained.

According to the embodiment described above, in the trajectory generation method, constraint conditions that the host vehicle 1 must satisfy while traveling along the trajectory are set. Then, a trajectory along which the host vehicle 1 travels is generated so as to satisfy the constraint conditions. The constraint conditions are set based on the probability of the state of the host vehicle 1 satisfying the predetermined conditions.

With this configuration, a trajectory is set to satisfy constraint conditions based on the probability of the state of the moving body (vehicle), and since the moving body is perceived with a probabilistic spread, it becomes easier to satisfy the constraint conditions, and it becomes possible to set a trajectory that does not impose excessive constraints on unlikely events, thereby increasing the opportunities for the vehicle to perform the desired action.

In addition, unless there are special restrictions, the order in which each process is performed may be changed.

Furthermore, the same effect can be achieved even if other configurations, examples of which are given in this specification, are appropriately added to the above configuration, i.e., other configurations in this specification that were not mentioned as the above configuration are appropriately added.

<Modifications of the above-described embodiments>
In the embodiments described above, the dimensions, shapes, relative positional relationships, and implementation conditions of each component may be described, but these are merely examples in all aspects and are not limiting.

Thus, numerous variations and equivalents not shown are contemplated within the scope of the technology disclosed herein, including, for example, the modification, addition, or omission of at least one component.

Furthermore, unless a contradiction arises, when it is stated in the embodiments described above that "one" component is provided, "one or more" of that component may be provided.

Furthermore, each component in the embodiments described above is a conceptual unit, and the scope of the technology disclosed in this specification includes cases where one component is made up of multiple structures, where one component corresponds to part of a structure, and even where multiple components are provided in one structure.

In addition, each component in the embodiments described above is intended to include structures having other structures or shapes so long as they perform the same function.

Furthermore, the descriptions in this specification are incorporated by reference for all purposes related to the present technology, and none of them are admitted to be prior art.

Various aspects of this disclosure are summarized below as appendices.

(Appendix 1)
a constraint condition setting unit for setting constraint conditions that a moving object passing through a trajectory must satisfy;
a trajectory generation unit for generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
The constraint condition includes a condition set based on a probability of a state of the moving object satisfying a predetermined condition.
Trajectory generator.

(Appendix 2)
2. A trajectory generation device according to claim 1,
The vehicle further includes an entry-prohibited area setting unit for setting an entry-prohibited area into which the moving object is prohibited from entering,
the constraint condition is set based on a probability that the moving object is outside the no-entry area;
Trajectory generator.

(Appendix 3)
3. A trajectory generation device according to claim 2,
An obstacle movement prediction unit is further provided for probabilistically predicting a state of an obstacle,
the no-entry area setting unit sets the no-entry area based on the state of the obstacle probabilistically predicted by the obstacle movement prediction unit;
Trajectory generator.

(Appendix 4)
4. A trajectory generation device according to any one of claims 1 to 3,
the constraint condition setting unit transforms a term indicating a probability of a state of the moving object in a conditional expression indicating the constraint condition into a term not including a probability;
Trajectory generator.

(Appendix 5)
5. A trajectory generation device according to any one of claims 1 to 4,
the constraint condition setting unit sets a probability of a state of the moving object under the constraint condition based on a time difference between a current time and a time when the moving object passes through the trajectory generated by the trajectory generating unit;
Trajectory generator.

(Appendix 6)
6. A trajectory generation device according to claim 5,
the constraint condition setting unit sets a lower probability of the state of the moving object under the constraint condition as the time difference between a current time and a time when the moving object passes through the trajectory generated by the trajectory generating unit increases;
Trajectory generator.

(Appendix 7)
7. A trajectory generation device according to any one of claims 1 to 6,
the constraint condition setting unit sets, in the constraint condition, a probability of a state of the moving object at a position on the trajectory generated by the trajectory generation unit based on a distance between the position on the trajectory and a current position of the moving object;
Trajectory generator.

(Appendix 8)
8. A trajectory generation device according to claim 7,
the constraint condition setting unit sets a lower probability of the state of the moving body at the position on the orbit as the distance between the position on the orbit and the current position of the moving body increases;
Trajectory generator.

(Appendix 9)
9. A trajectory generation device according to any one of claims 1 to 8,
the constraint condition setting unit sets a probability of the state of the moving object under the constraint condition based on at least one of a current action type of the moving object and a next action type of the moving object;
Trajectory generator.

(Appendix 10)
10. A trajectory generation device according to claim 9,
the constraint condition setting unit sets a higher probability of the state of the moving body under the constraint condition when the current action type of the moving body and the next action type to be performed by the moving body are different than when the two are the same;
Trajectory generator.

(Appendix 11)
11. A trajectory generation device according to claim 9 or 10,
the constraint condition setting unit sets a probability of a state of the moving object under the constraint condition based on a degree of necessity of the action type;
Trajectory generator.

(Appendix 12)
12. A trajectory generation device according to claim 11,
the constraint condition setting unit sets the degree of necessity of the behavior type based on a relationship between a trajectory structure at a position on the trajectory generated by the trajectory generating unit and the behavior type of the moving object;
Trajectory generator.

(Appendix 13)
13. A trajectory generation device according to any one of claims 1 to 12,
The trajectory generating unit:
When the trajectory that satisfies the constraint condition is obtained, it is determined that the next action type to be performed by the moving object is executable;
if the trajectory that satisfies the constraint condition is not obtained, it is determined that the next action type to be performed by the moving object is not executable.
Trajectory generator.

(Appendix 14)
Set constraints that must be satisfied by the moving object along the trajectory,
generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
The constraint condition includes a condition set based on a probability of a state of the moving object satisfying a predetermined condition.
Trajectory generation method.

1 Vehicle, 2 Steering wheel, 3 Steering shaft, 4 Steering unit, 5 EPS motor, 6 Power train unit, 7 Brake unit, 11 Front camera, 12 Radar sensor, 13 GNSS sensor, 15 V2X receiver, 16 Yaw rate sensor, 17 Speed sensor, 18 Acceleration sensor, 20 Steering angle sensor, 21 Steering torque sensor, 100 Vehicle control device, 100A Vehicle control device, 100B Vehicle control device, 110 Trajectory generation device, 110A Trajectory generation device, 110B Trajectory generation device, 111 Obstacle movement prediction unit, 112 No entry area setting unit, 113 Constraint condition setting unit, 114 Trajectory generation unit, 120 Vehicle control unit, 200 Information acquisition unit, 210 Vehicle information acquisition unit, 220 Obstacle information acquisition unit, 230 Road information acquisition unit, 300 Controller unit, 310 Power train controller, 320 brake controller, 330 EPS controller, 500 vehicle control system, 500A vehicle control system, 500B vehicle control system.

Claims (17)

a constraint condition setting unit for setting constraint conditions that a moving object passing through the trajectory must satisfy;
a trajectory generation unit for generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
the constraint condition setting unit sets a probability of a state of the moving object under the constraint condition based on a time difference between a current time and a time when the moving object passes through the trajectory generated by the trajectory generating unit;
Trajectory generator. a constraint condition setting unit for setting constraint conditions that a moving object passing through a trajectory must satisfy;
a trajectory generation unit for generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
the constraint condition setting unit sets, in the constraint condition, a probability of a state of the moving object at a position on the trajectory generated by the trajectory generation unit based on a distance between the position on the trajectory and a current position of the moving object;
Trajectory generator. a constraint condition setting unit for setting constraint conditions that a moving object passing through a trajectory must satisfy;
a trajectory generation unit for generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
the constraint condition setting unit sets a probability of the state of the moving object under the constraint condition based on at least one of a current action type of the moving object and the action type that the moving object will next perform;
the constraint condition setting unit sets a higher probability of the state of the moving body under the constraint condition when the current action type of the moving body and the next action type to be performed by the moving body are different than when the two are the same;
Trajectory generator. a constraint condition setting unit for setting constraint conditions that a moving object passing through a trajectory must satisfy;
a trajectory generation unit for generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
the constraint condition setting unit sets a probability of the state of the moving object under the constraint condition based on at least one of a current action type of the moving object and the action type that the moving object will next perform;
the constraint condition setting unit sets a probability of a state of the moving object under the constraint condition based on a degree of necessity of the action type;
Trajectory generator. a constraint condition setting unit for setting constraint conditions that a moving object passing through a trajectory must satisfy;
a trajectory generation unit for generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
the constraint condition setting unit transforms a term indicating a probability of a state of the moving object in a conditional expression indicating the constraint condition into a term not including a probability;
Trajectory generator. A trajectory generation device according to any one of claims 1 to 5 ,
The vehicle further includes an entry-prohibited area setting unit for setting an entry-prohibited area into which the moving object is prohibited from entering,
the constraint condition is set based on a probability that the moving object is outside the no-entry area;
Trajectory generator. The trajectory generating device according to claim 6 ,
An obstacle movement prediction unit is further provided for probabilistically predicting a state of an obstacle,
the no-entry area setting unit sets the no-entry area based on the state of the obstacle probabilistically predicted by the obstacle movement prediction unit;
Trajectory generator. A trajectory generation device according to any one of claims 1 to 4 ,
the constraint condition setting unit transforms a term indicating a probability of a state of the moving object in a conditional expression indicating the constraint condition into a term not including a probability;
Trajectory generator. The trajectory generating device according to claim 1 ,
the constraint condition setting unit sets a lower probability of the state of the moving object under the constraint condition as the time difference between a current time and a time when the moving object passes through the trajectory generated by the trajectory generating unit increases;
Trajectory generator. The trajectory generating device according to claim 2 ,
the constraint condition setting unit sets a lower probability of the state of the moving body at the position on the orbit as the distance between the position on the orbit and the current position of the moving body increases;
Trajectory generator. The trajectory generating device according to claim 4 ,
the constraint condition setting unit sets the degree of necessity of the behavior type based on a relationship between a trajectory structure at a position on the trajectory generated by the trajectory generating unit and the behavior type of the moving object;
Trajectory generator. A trajectory generation device according to any one of claims 1 to 5 ,
The trajectory generating unit:
When the trajectory that satisfies the constraint condition is obtained, it is determined that the next action type to be performed by the moving object is executable;
if the trajectory that satisfies the constraint condition is not obtained, it is determined that the next action type to be performed by the moving object is not executable.
Trajectory generator. Set constraints that must be satisfied by the moving object along the trajectory,
generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
setting the constraint condition includes setting a probability of a state of the moving object under the constraint condition based on a time difference between a current time and a time when the moving object passes through the trajectory;
Trajectory generation method. Set constraints that must be satisfied by the moving object along the trajectory,
generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
and setting the constraint condition includes setting, in the constraint condition, a probability of a state of the moving object at a position on the orbit based on a distance between the position on the orbit and a current position of the moving object.
Trajectory generation method. Set constraints that must be satisfied by the moving object along the trajectory,
generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
setting the constraint condition includes setting a probability of the state of the moving object under the constraint condition based on at least one of a current action type of the moving object and the action type that the moving object will next perform;
setting the constraint condition to a higher probability of the state of the moving body under the constraint condition when the current action type of the moving body and the next action type to be performed by the moving body are different than when the two are the same;
Trajectory generation method. Set constraints that must be satisfied by the moving object along the trajectory,
generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
setting the constraint condition includes setting a probability of the state of the moving object under the constraint condition based on at least one of a current action type of the moving object and the action type that the moving object will next perform;
setting the constraint condition includes setting a probability of a state of the moving object under the constraint condition based on a degree of necessity of the action type.
Trajectory generation method. Set constraints that must be satisfied by the moving object along the trajectory,
generating the trajectory along which the moving object passes so as to satisfy the constraint condition;
the constraint condition is set based on a probability of a state of the moving object satisfying a predetermined condition,
A term indicating a probability of a state of the moving object in a conditional expression indicating the constraint condition is transformed into a term not including a probability.
Trajectory generation method.