[go: up one dir, main page]

WO2019047657A1 - 无人驾驶车辆的数据训练方法和装置 - Google Patents

无人驾驶车辆的数据训练方法和装置 Download PDF

Info

Publication number
WO2019047657A1
WO2019047657A1 PCT/CN2018/099173 CN2018099173W WO2019047657A1 WO 2019047657 A1 WO2019047657 A1 WO 2019047657A1 CN 2018099173 W CN2018099173 W CN 2018099173W WO 2019047657 A1 WO2019047657 A1 WO 2019047657A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
vehicle
mode
driverless
unmanned vehicle
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2018/099173
Other languages
English (en)
French (fr)
Inventor
郁浩
闫泳杉
郑超
唐坤
张云飞
姜雨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Baidu Online Network Technology Beijing Co Ltd
Original Assignee
Baidu Online Network Technology Beijing Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Baidu Online Network Technology Beijing Co Ltd filed Critical Baidu Online Network Technology Beijing Co Ltd
Priority to EP18853701.3A priority Critical patent/EP3582056B1/en
Publication of WO2019047657A1 publication Critical patent/WO2019047657A1/zh
Priority to US16/566,826 priority patent/US20200004249A1/en
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0231—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means
    • G05D1/0242—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means using non-visible light signals, e.g. IR or UV signals
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
    • B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
    • B60W40/09—Driving style or behaviour
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W60/00—Drive control systems specially adapted for autonomous road vehicles
    • B60W60/005—Handover processes
    • B60W60/0051—Handover processes from occupants to vehicle
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W60/00—Drive control systems specially adapted for autonomous road vehicles
    • B60W60/005—Handover processes
    • B60W60/0053—Handover processes from vehicle to occupant
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
    • G05D1/0214—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory in accordance with safety or protection criteria, e.g. avoiding hazardous areas
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0259—Control of position or course in two dimensions specially adapted to land vehicles using magnetic or electromagnetic means
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0276—Control of position or course in two dimensions specially adapted to land vehicles using signals provided by a source external to the vehicle
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0276—Control of position or course in two dimensions specially adapted to land vehicles using signals provided by a source external to the vehicle
    • G05D1/028—Control of position or course in two dimensions specially adapted to land vehicles using signals provided by a source external to the vehicle using a RF signal
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
    • B60W2050/0062—Adapting control system settings
    • B60W2050/0075—Automatic parameter input, automatic initialising or calibrating means
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
    • B60W2050/0062—Adapting control system settings
    • B60W2050/0075—Automatic parameter input, automatic initialising or calibrating means
    • B60W2050/0083—Setting, resetting, calibration
    • B60W2050/0088—Adaptive recalibration
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2540/00—Input parameters relating to occupants
    • B60W2540/30—Driving style
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2556/00—Input parameters relating to data
    • B60W2556/10—Historical data

Definitions

  • the present application relates to the field of computer technology, and in particular to the technical field of driverless vehicles, and more particularly to a data training method and apparatus for an unmanned vehicle.
  • the driverless vehicle is a new type of intelligent car. It mainly performs precise control and calculation analysis on various parts of the vehicle through vehicular terminal equipment such as ECU (Electronic Control Unit) to realize the fully automatic operation of the vehicle.
  • vehicular terminal equipment such as ECU (Electronic Control Unit) to realize the fully automatic operation of the vehicle.
  • the purpose of the vehicle being unmanned.
  • the vehicle-mounted terminal device is usually trained by using a machine learning method. Therefore, the collection of training data is of great significance for the safe driving of an unmanned vehicle.
  • Unmanned vehicles can execute vehicle control commands to achieve accurate operation of unmanned driving.
  • the actual road conditions usually do not have strict and accurate information, which causes the unmanned vehicle to deviate from the actual road conditions when the driver is unmanned according to the vehicle control command.
  • the deviation reaches a certain level, some parameters measured by the driverless vehicle are out of the preset range, thereby making the unmanned vehicle unable to execute the vehicle control command.
  • it is usually necessary to enter the manual driving mode to control the vehicle to return to the normal driving state by the driver.
  • the situation in which an unmanned vehicle cannot execute a vehicle control command is highly random, and it is often difficult to collect possible deviations by conventional means.
  • the artificial driving mode also has similar uncertainties, and it is also difficult to collect corresponding data, which reduces the safety of driverless driving.
  • the purpose of the embodiments of the present application is to provide a data training method and apparatus for an unmanned vehicle to solve the technical problems mentioned in the above background art.
  • an embodiment of the present application provides a data training method for an unmanned vehicle, which acquires sensor data and a modified sample of an unmanned vehicle, wherein the modified sample is used to represent an unmanned vehicle during a driving process.
  • an end-to-end model is constructed by the sensor data and the modified samples for outputting control commands corresponding to the driver's driving behavior through the sensor data and the corrected samples.
  • the method further comprises: pushing the end-to-end model to the driverless vehicle and correcting the end-to-end model with the measured feedback data.
  • the method for acquiring the modified sample is: acquiring vehicle data of the driverless vehicle; determining, based on the vehicle data, whether the unmanned vehicle is switched from the driverless mode to the manual driving mode, and then manually Driving mode is switched to an unmanned mode; in response to determining that the unmanned vehicle is switched from the driverless mode to the manual driving mode, determining a first switching moment of switching to the manual driving mode, in response to determining that the unmanned vehicle is manually The driving mode is switched to the driverless mode, and the second switching timing of switching to the driverless mode is determined; the vehicle data acquired between the first switching moment and the second switching moment is marked as a correction sample.
  • the method further includes marking the vehicle data acquired within the preset time period before the first switching moment as a negative sample, the negative sample being used to characterize interference encountered by the unmanned vehicle.
  • an embodiment of the present application provides a data training device for an unmanned vehicle, where the device includes: a sample acquisition unit, configured to acquire sensor data and a modified sample of an unmanned vehicle, wherein the modified sample a driving behavior data for characterizing a driver when an unmanned vehicle encounters an obstacle during driving; a data training unit for constructing an end-to-end model by using the sensor data and the modified sample, the end-to-end model being used for A control command corresponding to the driver's driving behavior is output through the sensor data and the corrected sample.
  • the apparatus includes a correction unit for pushing the end-to-end model to the driverless vehicle and correcting the end-to-end model with the measured feedback data.
  • the apparatus further includes: a modified sample collection unit, configured to acquire a modified sample, the modified sample collection unit comprising: a vehicle data acquisition subunit, configured to acquire vehicle data of the driverless vehicle; and a driving mode a determining subunit, configured to determine whether the unmanned vehicle is switched from the driverless mode to the manual driving mode based on the vehicle data, and then switch from the manual driving mode to the driverless mode; and switch the timing determining subunit for responding Determining that the unmanned vehicle is switched from the driverless mode to the manual driving mode, determining a first switching moment of switching to the manual driving mode, and determining to change the unmanned vehicle from the manual driving mode to the driverless mode, determining Switching to the second switching moment of the driverless mode; the first marking subunit is configured to mark the vehicle data acquired between the first switching moment and the second switching moment as a modified sample.
  • a modified sample collection unit configured to acquire a modified sample
  • the modified sample collection unit comprising: a vehicle data acquisition subunit, configured to acquire vehicle data of the driverless vehicle;
  • the vehicle data includes sensor data; and the driving mode determining subunit includes: a first determining module, configured to determine whether pressure data collected by a pressure sensor on a steering wheel of the unmanned vehicle is greater than a preset And a second determining module, configured to determine whether the temperature data collected by the temperature sensor on the steering wheel of the unmanned vehicle is greater than a preset temperature threshold; and the first determining module is configured to use the pressure data If the preset pressure threshold is greater than the preset pressure threshold and/or the temperature data is greater than the preset temperature threshold, it is determined that the unmanned vehicle is switched from the driverless mode to the manual driving mode.
  • the vehicle data includes expected driving data and actual driving data; and the driving mode determining subunit includes: a third determining module, configured to determine whether a difference between the expected driving data and the actual driving data is The second determining module is configured to determine that the unmanned vehicle is switched from the driverless mode to the manual driving mode if the difference is greater than a preset threshold.
  • the apparatus further includes: a second marking unit, configured to mark the vehicle data acquired within the preset time period before the first switching moment as a negative sample, and the negative sample is used to represent the unmanned vehicle encounter The disturbance to the arrival.
  • a second marking unit configured to mark the vehicle data acquired within the preset time period before the first switching moment as a negative sample, and the negative sample is used to represent the unmanned vehicle encounter The disturbance to the arrival.
  • an embodiment of the present application provides a server, including: one or more processors; and a memory, configured to store one or more programs, when the one or more programs are executed by the one or more processors.
  • the one or more processors are caused to perform the data training method of the unmanned vehicle of the first aspect described above.
  • an embodiment of the present application provides a computer readable storage medium, where a computer program is stored thereon, wherein the program is executed by a processor to implement the data training method of the unmanned vehicle of the first aspect. .
  • the data training method and apparatus for an unmanned vehicle provided by the embodiment of the present application first acquire sensor data and a modified sample of an unmanned vehicle, and then construct an end-to-end model by using the sensor data and the modified sample, thereby improving unmanned driving. Driving safety of the vehicle.
  • FIG. 1 is an exemplary system architecture diagram to which the present application can be applied;
  • FIG. 2 is a flow chart of one embodiment of a data training method for an unmanned vehicle according to the present application
  • FIG. 3 is a schematic diagram of an application scenario of a data training method for an unmanned vehicle according to the present application
  • FIG. 4 is a schematic structural view of an embodiment of a data training device for an unmanned vehicle according to the present application.
  • FIG. 5 is a schematic structural diagram of a computer system suitable for implementing a server of an embodiment of the present application.
  • FIG. 1 illustrates an exemplary system architecture 100 of an embodiment of a data training method of an unmanned vehicle or a data training device for an unmanned vehicle to which the present application may be applied.
  • system architecture 100 can include unmanned vehicles 101, 102, 103, network 104, and server 105.
  • the network 104 is used to provide a medium for communication links between the unmanned vehicles 101, 102, 103 and the server 105.
  • Network 104 may include various types of connections, such as wired, wireless communication links, fiber optic cables, and the like.
  • the driverless vehicles 101, 102, 103 interact with the server 105 over the network 104 to receive or transmit messages and the like.
  • Various electronic devices such as pressure sensors, temperature sensors, distance sensors, data memories, data transceivers, and the like can be mounted on the unmanned vehicles 101, 102, and 103.
  • the driverless vehicles 101, 102, 103 may be various vehicles having a plurality of sample acquisition units and data training units, including but not limited to electric vehicles, hybrid electric vehicles, and internal combustion engine vehicles.
  • the server 105 may be a server that provides various services, such as data processing of sensor data and correction samples collected by the unmanned vehicles 101, 102, 103 to obtain an end-to-end model server.
  • the server can analyze the sensor data and the modified samples collected by the unmanned vehicles 101, 102, and 103, determine the switching between the driverless mode and the manual driving mode of the unmanned vehicles 101, 102, and 103, and further obtain the end. To the end model.
  • the data training method of the driverless vehicle may be separately executed by the unmanned vehicles 101, 102, 103, or may also be performed by the unmanned vehicles 101, 102, 103 and the server 105. Co-execution. Accordingly, the data training device of the driverless vehicle may be disposed in the unmanned vehicle 101, 102, 103 or may be disposed in the server 105.
  • the data training method of the driverless vehicle includes the following steps:
  • Step 201 Acquire sensor data and a modified sample of the driverless vehicle.
  • the electronic device on which the data training method of the driverless vehicle runs (for example, the server 105 shown in FIG. 1 or the brain of the driverless vehicle on the unmanned vehicles 101, 102, 103) can pass
  • the vehicle data of the driverless vehicle is acquired from the unmanned vehicles 101, 102, 103 by a wired connection method or a wireless connection method.
  • the brain of the driverless vehicle can be an in-vehicle electronic terminal device with data processing functions.
  • the above wireless connection methods may include but are not limited to 3G/4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other currently known or future developed. Wireless connection method.
  • the control method of the driverless vehicles 101, 102, 103 is usually set in the case of normal running. Therefore, when the unmanned vehicles 101, 102, 103 encounter an emergency (ie, interference) during traveling, the unmanned vehicles 101, 102, 103 themselves are often unable to output a reasonable control command according to an emergency. At this time, the driver is required to switch the unmanned vehicles 101, 102, 103 from the driverless mode to the manual driving mode, and to avoid an emergency by the driver's driving behavior.
  • the above process can collect corresponding sensor data and correction samples. Wherein, the modified sample is used to represent driving behavior data of the driver when the unmanned vehicle 101, 102, 103 encounters interference during driving.
  • Step 202 constructing an end-to-end model by using the sensor data and the modified samples.
  • the sensor data and the corrected samples have a corresponding relationship.
  • the modified sample is used to characterize the driver's driving behavior data when the unmanned vehicle 101, 102, 103 encounters interference during driving, and the data of the emergency corresponding to the corrected sample can be found from the sensor data.
  • the end-to-end model can then be built by learning the sensor data and the modified samples.
  • the end-to-end model is used to output a control command corresponding to the driver's driving behavior through the sensor data and the corrected sample.
  • the method may further include: pushing the end-to-end model to the driverless vehicle, and correcting the end-to-end model by the measured feedback data.
  • the end-to-end model is introduced into the unmanned vehicles 101, 102, and 103, so that the unmanned vehicles 101, 102, and 103 can pass the end-to-end model in time when encountering similar emergencies.
  • Output corresponding control commands to avoid unexpected events. In practice, incidents may occur from different locations or in different forms.
  • feedback data it is also necessary to collect data collected by the sensor when the end-to-end model is applied on an unmanned vehicle, that is, feedback data.
  • the end-to-end model is modified by feedback data to improve the accuracy and robustness of the end-to-end model output control commands.
  • the method for collecting the modified sample is:
  • the first step is to obtain vehicle data for the driverless vehicle.
  • the electronic device on which the data training method of the driverless vehicle runs (for example, the server 105 shown in FIG. 1 or the brain of the driverless vehicle on the unmanned vehicles 101, 102, 103) can pass
  • the vehicle data of the driverless vehicle is acquired from the unmanned vehicles 101, 102, 103 by a wired connection method or a wireless connection method.
  • the brain of the driverless vehicle can be an in-vehicle electronic terminal device with data processing functions.
  • the above wireless connection methods may include but are not limited to 3G/4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other currently known or future developed. Wireless connection method.
  • the vehicle data may include data collected by various types of sensors installed on the unmanned vehicles 101, 102, 103, such as distance data between the unmanned vehicles 101, 102, 103 collected by the distance sensor and surrounding objects; angle sensors Angle data of the collected tire deflection direction; travel speed data of the unmanned vehicles 101, 102, and 103 collected by the speed sensor; operating temperature data of the engine collected by the temperature sensor, and the like.
  • the vehicle data may also include driving state data of the unmanned vehicles 101, 102, 103, for example, the unmanned vehicles 101, 102, 103 are currently in a constant traveling state, a straight traveling state, or the door is in an open state.
  • the second step based on the vehicle data described above, it is determined whether the unmanned vehicle is switched from the driverless mode to the manual driving mode, and then switched from the manual driving mode to the driverless mode.
  • the unmanned vehicles 101, 102, 103 may not be able to normally execute the vehicle control command, thereby It is prone to unsafe driving.
  • the manual driving mode is switched to the driverless mode. In this process, the situation in which the unmanned vehicles 101, 102, 103 cannot properly execute the vehicle control command can be considered as interference.
  • the manual driving mode process after the manual intervention can be considered as a process for eliminating interference.
  • the unmanned vehicles 101, 102, 103 are switched to the driverless mode.
  • the unmanned vehicles 101, 102, 103 can record the above-described processes through various sensors to obtain corresponding vehicle data.
  • the unmanned vehicles 101, 102, 103 can automatically travel according to the line that needs to be set.
  • the unmanned vehicle 101, 102, 103 is switched from the driverless mode to the manual driving mode, or when the manual driving mode is switched to the unmanned mode, the server 105 or the driverless vehicle brain can be the vehicle.
  • the data corresponding to the above mode switching is found in the data.
  • the present embodiment needs to determine whether the unmanned vehicles 101, 102, 103 are switched from the driverless mode to the manual driving mode, and then switch from the manual driving mode to the driverless mode.
  • the process of switching from the driverless mode to the manual driving mode and then switching from the manual driving mode to the driverless mode can be considered as a process in which interference occurs and interference is eliminated. Collecting vehicle data for this process can help to improve the driving safety of the unmanned vehicles 101, 102, 103.
  • the driver In the manual driving mode, the driver is required to control the steering wheel, and at this time, pressure is applied to the steering wheel. In this way, it is possible to monitor whether or not the manual driving mode is performed by a pressure sensor provided on the steering wheel. Similarly, in the manual driving mode, the driver's hand needs to touch the steering wheel. At this time, it is also possible to detect that the temperature of the partial area on the steering wheel is different from the other areas by the temperature sensor provided on the steering wheel, thereby monitoring that the unmanned vehicles 101, 102, and 103 are in the manual driving mode.
  • the driverless vehicles 101, 102, 103 can perform unmanned driving in accordance with the expected driving data, and simultaneously record the actual driving data of the unmanned vehicles 101, 102, 103. Although it may be subject to external interference, the unmanned vehicles 101, 102, 103 may be considered to be in normal unmanned driving as long as the difference between the expected driving data and the actual driving data is within a certain error range. Conversely, when the difference between the expected driving data and the actual driving data exceeds a preset threshold, it can be considered that the driving state of the driverless vehicles 101, 102, 103 is switched from the driverless mode to the manual driving mode.
  • determining a first switching moment of switching to the manual driving mode in response to determining that the unmanned vehicle is switched from the manual driving mode to none
  • the human driving mode determines the second switching moment to switch to the driverless mode.
  • the corresponding switching timing can be found from the vehicle data.
  • the timing at which the driverless vehicles 101, 102, 103 are switched from the driverless mode to the manual driving mode is determined as the first switching moment; the unmanned vehicles 101, 102, 103 are switched from the manual driving mode to The moment of the driverless mode is determined as the second switching moment.
  • the vehicle data acquired between the first switching moment and the second switching moment is marked as a modified sample.
  • the first switching moment is a moment when the unmanned vehicles 101, 102, 103 are switched from the driverless mode to the manual driving mode; and the second switching moment is that the driverless vehicles 101, 102, 103 are switched from the manual driving mode to the driverless driving mode.
  • the moment of the pattern is a moment.
  • the unmanned vehicles 101, 102, 103 are in the manual driving mode.
  • the process of the manual driving mode is an anti-interference process. Therefore, the vehicle data acquired between the first switching timing and the second switching timing can be marked as a corrected sample.
  • the method of the embodiment may further include: marking the vehicle data acquired in the preset time period before the first switching time as a negative sample.
  • the unmanned vehicles 101, 102, 103 need to switch from the driverless mode to the manual driving mode when they are not normally in the driverless mode. It can be considered that the unmanned vehicles 101, 102, 103 are disturbed (or encounter an emergency) for a period of time before switching to the manual driving mode. Therefore, the vehicle data acquired within the preset time period before the first switching timing can be marked as a negative sample.
  • the negative samples of this embodiment are used to characterize the interference experienced by the unmanned vehicles 101, 102, 103. Among them, the negative sample is used to characterize the interference encountered by the unmanned vehicle, and the specific value of the preset time period is determined according to actual needs.
  • the method of the embodiment may further include: training the preset unmanned vehicle brain model by using the modified sample and the negative sample, wherein the brain model of the unmanned vehicle is used for The vehicle control command is predicted based on the vehicle data.
  • the negative sample and the modified sample can be used as training samples to train the brain model of the driverless vehicle, so that the brain model of the driverless vehicle can be learned from the modified sample, and how to issue the vehicle control command in the case of a negative sample. Eliminate interference caused by negative samples.
  • FIG. 3 is a schematic diagram of an application scenario of a data training method for an unmanned vehicle according to the present embodiment.
  • the server 105 acquires sensor data and correction samples of the driverless vehicle, and then constructs an end-to-end model through the sensor data and the modified samples, so that the driverless vehicle can encounter similar interference when encountering similar interference. Output the correct control commands in time.
  • the method provided by the above embodiment of the present application improves the driving safety of an unmanned vehicle.
  • the present application provides an embodiment of a data training device for an unmanned vehicle, the device embodiment corresponding to the method embodiment shown in FIG.
  • the device can be specifically applied to various electronic devices.
  • the data training device 400 of the driverless vehicle of the present embodiment may include a sample acquisition unit 401 and a data training unit 402.
  • the sample obtaining unit 401 is configured to acquire sensor data and a modified sample of the driverless vehicle, wherein the modified sample is used to represent driving behavior data of the driver when the unmanned vehicle encounters interference during driving;
  • the training unit 402 is configured to construct an end-to-end model by using the sensor data and the modified sample, and the end-to-end model is configured to output a control instruction corresponding to the driving behavior of the driver through the sensor data and the corrected sample.
  • the apparatus may include: a correction unit (not shown) for pushing the end-to-end model to the unmanned vehicle and passing the measured The feedback data corrects the end-to-end model.
  • the apparatus may include: a modified sample collection unit (not shown) for acquiring a modified sample
  • the modified sample collection unit may include: a vehicle data acquisition sub- A unit (not shown), a driving mode judging subunit (not shown), a switching timing determining subunit (not shown), and a first marking subunit (not shown).
  • the vehicle data acquisition subunit is configured to acquire vehicle data of the driverless vehicle;
  • the driving mode determination subunit is configured to determine whether the unmanned vehicle is switched from the driverless mode to the manual driving mode based on the vehicle data, and then The manual driving mode is switched to the driverless mode;
  • the switching timing determining subunit is configured to determine a first switching moment to switch to the manual driving mode in response to determining that the unmanned vehicle switches from the driverless mode to the manual driving mode, in response to Determining that the unmanned vehicle is switched from the manual driving mode to the driverless mode, and determining a second switching moment to switch to the driverless mode;
  • the first marking subunit is configured to use the first switching moment and the second switching moment The acquired vehicle data is marked as a modified sample.
  • the vehicle data may include sensor data; and the driving mode determining subunit may include: a first determining module (not shown) and a second determining module (in the figure) Not shown) and the first determination module (not shown).
  • the first determining module is configured to determine whether the pressure data collected by the pressure sensor on the steering wheel of the unmanned vehicle is greater than a preset pressure threshold; and/or the second determining module is configured to determine the unmanned vehicle Whether the temperature data collected by the temperature sensor on the steering wheel is greater than a preset temperature threshold; the first determining module is configured to determine the above if the pressure data is greater than a preset pressure threshold and/or the temperature data is greater than a preset temperature threshold
  • the driverless vehicle switches from the driverless mode to the manual driving mode.
  • the vehicle data may include expected driving data and actual driving data; and the driving mode determining subunit may include: a third determining module (not shown) and a second Determine the module (not shown).
  • the third determining module is configured to determine whether a difference between the expected driving data and the actual driving data is greater than a preset threshold; and the second determining module is configured to determine that the difference is greater if the difference is greater than a preset threshold. The person driving the vehicle switches from the driverless mode to the manual driving mode.
  • the data training device 400 of the driverless vehicle may further include: a second marking unit (not shown) for presetting the preset time period before the first switching moment
  • the internally acquired vehicle data is labeled as a negative sample that is used to characterize the interference encountered by the unmanned vehicle.
  • the embodiment further provides a server comprising: one or more processors; a memory for storing one or more programs, when the one or more programs are executed by the one or more processors, One or more processors execute the data training method of the above-described unmanned vehicle.
  • the embodiment further provides a computer readable storage medium having stored thereon a computer program that, when executed by the processor, implements the data training method of the above-described unmanned vehicle.
  • FIG. 5 there is shown a block diagram of a computer system 500 suitable for use in implementing the server of the embodiments of the present application.
  • the server shown in FIG. 5 is merely an example, and should not impose any limitation on the function and scope of use of the embodiments of the present application.
  • computer system 500 includes a central processing unit (CPU) 501 that can be loaded into a program in random access memory (RAM) 503 according to a program stored in read only memory (ROM) 502 or from storage portion 508. And perform various appropriate actions and processes.
  • RAM random access memory
  • ROM read only memory
  • RAM 503 various programs and data required for the operation of the system 500 are also stored.
  • the CPU 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504.
  • An input/output (I/O) interface 505 is also coupled to bus 504.
  • the following components are connected to the I/O interface 505: an input portion 506 including a keyboard, a mouse, etc.; an output portion 507 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), and the like, and a storage portion 508 including a hard disk or the like. And a communication portion 509 including a network interface card such as a LAN card, a modem, or the like. The communication section 509 performs communication processing via a network such as the Internet.
  • Driver 510 is also coupled to I/O interface 505 as needed.
  • a removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory or the like is mounted on the drive 510 as needed so that a computer program read therefrom is installed into the storage portion 508 as needed.
  • an embodiment of the present disclosure includes a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for executing the method illustrated in the flowchart.
  • the computer program can be downloaded and installed from the network via the communication portion 509, and/or installed from the removable medium 511.
  • CPU central processing unit
  • the computer readable medium described above may be a computer readable signal medium or a computer readable storage medium or any combination of the two.
  • the computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the foregoing.
  • a computer readable storage medium may be any tangible medium that can contain or store a program, which can be used by or in connection with an instruction execution system, apparatus or device.
  • a computer readable signal medium may include a data signal that is propagated in the baseband or as part of a carrier, carrying computer readable program code. Such propagated data signals can take a variety of forms including, but not limited to, electromagnetic signals, optical signals, or any suitable combination of the foregoing.
  • the computer readable signal medium can also be any computer readable medium other than a computer readable storage medium, which can transmit, propagate, or transport a program for use by or in connection with the instruction execution system, apparatus, or device.
  • Program code embodied on a computer readable medium can be transmitted by any suitable medium, including but not limited to wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
  • each block of the flowchart or block diagram can represent a module, a program segment, or a portion of code that includes one or more of the logic functions for implementing the specified.
  • Executable instructions can also occur in a different order than that illustrated in the drawings. For example, two successively represented blocks may in fact be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts can be implemented in a dedicated hardware-based system that performs the specified function or operation. Or it can be implemented by a combination of dedicated hardware and computer instructions.
  • the units involved in the embodiments of the present application may be implemented by software or by hardware.
  • the described unit may also be provided in the processor, for example, as a processor comprising a sample acquisition unit and a data training unit.
  • the names of these units do not constitute a limitation on the unit itself under certain circumstances.
  • the data training unit can also be described as "a unit for training an end-to-end model.”
  • the present application also provides a computer readable medium, which may be included in the apparatus described in the above embodiments, or may be separately present and not incorporated into the apparatus.
  • the computer readable medium carries one or more programs that, when executed by the device, cause the device to: acquire sensor data and a modified sample of the driverless vehicle, wherein the modified sample is for Characterizing the driver's driving behavior data when the unmanned vehicle encounters interference during driving; constructing an end-to-end model by using the sensor data and the modified sample, the end-to-end model is used to correspond to the sensor data and the corrected sample output Control instructions for the driver's driving behavior.

Landscapes

  • Engineering & Computer Science (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Mathematical Physics (AREA)
  • Electromagnetism (AREA)
  • Human Computer Interaction (AREA)
  • Traffic Control Systems (AREA)
  • Business, Economics & Management (AREA)
  • Medical Informatics (AREA)
  • Game Theory and Decision Science (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)

Abstract

本申请实施例公开了无人驾驶车辆的数据训练方法和装置。该方法的一具体实施方式包括：获取无人驾驶车辆的传感器数据和修正样本，其中，所述修正样本用于表征无人驾驶车辆行驶过程中遇到干扰时，驾驶员的驾驶行为数据；通过所述传感器数据和修正样本构建端到端模型，所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。该实施方式提高了无人驾驶车辆的行驶安全性。

Description

无人驾驶车辆的数据训练方法和装置
相关申请的交叉引用
本专利申请要求于2017年9月5日提交的、申请号为201710790859.0、申请人为百度在线网络技术(北京)有限公司、发明名称为“无人驾驶车辆的数据训练方法和装置”的中国专利申请的优先权，该申请的全文以引用的方式并入本申请中。
技术领域
本申请涉及计算机技术领域，具体涉及无人驾驶车辆技术领域，尤其涉及无人驾驶车辆的数据训练方法和装置。
背景技术
无人驾驶车辆是一种新型的智能汽车，主要通过ECU(Electronic Control Unit，电子控制单元)等车载终端设备对车辆中各个部分进行精准的控制与计算分析，从而实现车辆的全自动运行，达到车辆无人驾驶的目的。现有技术中，通常利用机器学习方法对车载终端设备进行训练，因此，训练数据的采集对无人驾驶车辆的安全驾驶而言具有非常重要的意义。
无人驾驶车辆可以执行车辆控制指令，实现对无人驾驶的准确操作。但实际中的路况通常不具有严格准确的信息，这就导致无人驾驶车辆根据车辆控制指令进行无人驾驶时，与实际的路况出现偏差。当这种偏差达到一定程度时，会使得无人驾驶车辆测得的某些参数超出预设范围，进而使得无人驾驶车辆无法执行车辆控制指令。当这种情况发生时，通常需要进入人工驾驶模式，通过驾驶员来控制车辆恢复到正常的行驶状态。而无人驾驶车辆无法执行车辆控制指令的情况随机性很强，通常不易通过常规的手段采集到可能的偏差。同时，人工驾驶模式也具有类似的不确定性，同样不易采集到对应的数据，这就 降低了无人驾驶的安全性。
发明内容
本申请实施例的目的在于提出了无人驾驶车辆的数据训练方法和装置，来解决以上背景技术部分提到的技术问题。
第一方面，本申请实施例提供了一种无人驾驶车辆的数据训练方法，获取无人驾驶车辆的传感器数据和修正样本，其中，所述修正样本用于表征无人驾驶车辆行驶过程中遇到干扰时，驾驶员的驾驶行为数据；通过所述传感器数据和修正样本构建端到端模型，所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。
在一些实施例中，所述方法还包括：将所述端到端模型推送给无人驾驶车辆，并通过测得的反馈数据对所述端到端模型进行修正。
在一些实施例中，所述修正样本的采集方法为：获取无人驾驶车辆的车辆数据；基于上述车辆数据，判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，之后从人工驾驶模式切换到无人驾驶模式；响应于确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式，确定切换到人工驾驶模式的第一切换时刻，响应于确定上述无人驾驶车辆从人工驾驶模式切换到无人驾驶模式，确定切换到无人驾驶模式的第二切换时刻；将上述第一切换时刻和第二切换时刻之间获取的车辆数据标记为修正样本。
在一些实施例中，上述车辆数据包括传感器数据；以及上述判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，包括：判断上述无人驾驶车辆的方向盘上的压力传感器所采集的压力数据是否大于预设的压力阈值；和/或判断上述无人驾驶车辆的方向盘上的温度传感器所采集的温度数据是否大于预设的温度阈值；若是，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
在一些实施例中，上述车辆数据包括预期驾驶数据和实际驾驶数据；以及上述判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，包括：判断上述预期驾驶数据与上述实际驾驶数据之间的 差值是否大于预设的阈值；若是，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
在一些实施例中，上述方法还包括：将上述第一切换时刻之前预设时段内获取的车辆数据标记为负样本，所述负样本用于表征无人驾驶车辆遇到的干扰。
第二方面，本申请实施例提供了一种无人驾驶车辆的数据训练装置，上述该装置包括：样本获取单元，用于获取无人驾驶车辆的传感器数据和修正样本，其中，所述修正样本用于表征无人驾驶车辆行驶过程中遇到干扰时，驾驶员的驾驶行为数据；数据训练单元，用于通过所述传感器数据和修正样本构建端到端模型，所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。
在一些实施例中，所述装置包括：修正单元，用于将所述端到端模型推送给无人驾驶车辆，并通过测得的反馈数据对所述端到端模型进行修正。
在一些实施例中，所述装置还包括：修正样本采集单元，用于采集修正样本，所述修正样本采集单元包括：车辆数据获取子单元，用于获取无人驾驶车辆的车辆数据；驾驶模式判断子单元，用于基于上述车辆数据，判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，之后从人工驾驶模式切换到无人驾驶模式；切换时刻确定子单元，用于响应于确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式，确定切换到人工驾驶模式的第一切换时刻，响应于确定上述无人驾驶车辆从人工驾驶模式切换到无人驾驶模式，确定切换到无人驾驶模式的第二切换时刻；第一标记子单元，用于将上述第一切换时刻和第二切换时刻之间获取的车辆数据标记为修正样本。
在一些实施例中，上述车辆数据包括传感器数据；以及上述驾驶模式判断子单元包括：第一判断模块，用于判断上述无人驾驶车辆的方向盘上的压力传感器所采集的压力数据是否大于预设的压力阈值；和/或第二判断模块，用于判断上述无人驾驶车辆的方向盘上的温度传感器所采集的温度数据是否大于预设的温度阈值；第一确定模块，用于若上述压力数据大于预设的压力阈值和/或上述温度数据大于预设 的温度阈值，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
在一些实施例中，上述车辆数据包括预期驾驶数据和实际驾驶数据；以及上述驾驶模式判断子单元包括：第三判断模块，用于判断上述预期驾驶数据与上述实际驾驶数据之间的差值是否大于预设的阈值；第二确定模块，用于若上述差值大于预设的阈值，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
在一些实施例中，上述装置还包括：第二标记单元，用于将上述第一切换时刻之前预设时段内获取的车辆数据标记为负样本，所述负样本用于表征无人驾驶车辆遇到的干扰。
第三方面，本申请实施例提供了一种服务器，包括：一个或多个处理器；存储器，用于存储一个或多个程序，当上述一个或多个程序被上述一个或多个处理器执行时，使得上述一个或多个处理器执行上述第一方面的无人驾驶车辆的数据训练方法。
第四方面，本申请实施例提供了一种计算机可读存储介质，其上存储有计算机程序，其特征在于，该程序被处理器执行时实现上述第一方面的无人驾驶车辆的数据训练方法。
本申请实施例提供的无人驾驶车辆的数据训练方法和装置，首先获取无人驾驶车辆的传感器数据和修正样本，之后通过所述传感器数据和修正样本构建端到端模型，提高了无人驾驶车辆的行驶安全性。
附图说明
通过阅读参照以下附图所作的对非限制性实施例所作的详细描述，本申请的其它特征、目的和优点将会变得更明显：
图1是本申请可以应用于其中的示例性系统架构图；
图2是根据本申请的无人驾驶车辆的数据训练方法的一个实施例的流程图；
图3是根据本申请的无人驾驶车辆的数据训练方法的一个应用场景的示意图；
图4是根据本申请的无人驾驶车辆的数据训练装置的一个实施例 的结构示意图；
图5是适于用来实现本申请实施例的服务器的计算机系统的结构示意图。
具体实施方式
下面结合附图和实施例对本申请作进一步的详细说明。可以理解的是，此处所描述的具体实施例仅仅用于解释相关发明，而非对该发明的限定。另外还需要说明的是，为了便于描述，附图中仅示出了与有关发明相关的部分。
需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本申请。
图1示出了可以应用本申请的无人驾驶车辆的数据训练方法或无人驾驶车辆的数据训练装置的实施例的示例性系统架构100。
如图1所示，系统架构100可以包无人驾驶车辆101、102、103，网络104和服务器105。网络104用以在无人驾驶车辆101、102、103和服务器105之间提供通信链路的介质。网络104可以包括各种连接类型，例如有线、无线通信链路或者光纤电缆等等。
无人驾驶车辆101、102、103通过网络104与服务器105交互，以接收或发送消息等。无人驾驶车辆101、102、103上可以安装有各种电子设备，例如，压力传感器、温度传感器、距离传感器、数据存储器、数据收发器等。
无人驾驶车辆101、102、103可以是具有多个样本获取单元和数据训练单元的各种车辆，包括但不限于电动汽车、油电混合汽车和内燃机汽车等等。
服务器105可以是提供各种服务的服务器，例如对无人驾驶车辆101、102、103采集的传感器数据和修正样本进行数据处理，以得到端到端模型的服务器。服务器可以对无人驾驶车辆101、102、103采集的传感器数据和修正样本进行分析，确定无人驾驶车辆101、102、103的无人驾驶模式和人工驾驶模式之间切换情况，并进而得到端到端模 型。
需要说明的是，本申请实施例所提供的无人驾驶车辆的数据训练方法可以由无人驾驶车辆101、102、103单独执行，或者也可以由无人驾驶车辆101、102、103和服务器105共同执行。相应地，无人驾驶车辆的数据训练装置可以设置于无人驾驶车辆101、102、103中，也可以设置于服务器105中。
应该理解，图1中的无人驾驶车辆、网络和服务器的数目仅仅是示意性的。根据实现需要，可以具有任意数目的无人驾驶车辆、网络和服务器。
继续参考图2，示出了根据本申请的无人驾驶车辆的数据训练方法的一个实施例的流程200。该无人驾驶车辆的数据训练方法包括以下步骤：
步骤201，获取无人驾驶车辆的传感器数据和修正样本。
在本实施例中，无人驾驶车辆的数据训练方法运行于其上的电子设备(例如图1所示的服务器105或无人驾驶车辆101、102、103上的无人驾驶车辆大脑)可以通过有线连接方式或者无线连接方式从无人驾驶车辆101、102、103获取无人驾驶车辆的车辆数据。无人驾驶车辆大脑可以是具有数据处理功能的车载电子终端设备。实践中，需要指出的是，上述无线连接方式可以包括但不限于3G/4G连接、WiFi连接、蓝牙连接、WiMAX连接、Zigbee连接、UWB(ultra wideband)连接、以及其他现在已知或将来开发的无线连接方式。
由于无人驾驶车辆101、102、103的控制方法通常是在正常行驶时的情况下设定的。因此，当无人驾驶车辆101、102、103在行驶过程中遇到突发事件(即干扰)时，无人驾驶车辆101、102、103自身往往不能根据突发事件输出合理的控制指令。此时，需要驾驶员将无人驾驶车辆101、102、103从无人驾驶模式切换到人工驾驶模式，并通过驾驶员的驾驶行为规避突发事件。上述的过程可以采集到对应的传感器数据和修正样本。其中，所述修正样本用于表征无人驾驶车辆101、102、103行驶过程中遇到干扰时，驾驶员的驾驶行为数据。
步骤202，通过所述传感器数据和修正样本构建端到端模型。
由上述描述可知，上述的传感器数据和修正样本具有对应关系。修正样本用于表征无人驾驶车辆101、102、103行驶过程中遇到干扰时，驾驶员的驾驶行为数据，则可以从传感器数据中找到与修正样本对应的突发事件的数据。之后，可以通过对传感器数据和修正样本的学习，来构建端到端模型。所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。
在本实施例的一些可选的实现方式中，所述方法还可以包括：将所述端到端模型推送给无人驾驶车辆，并通过测得的反馈数据对所述端到端模型进行修正。
得到端到端模型后，将端到端模型导入无人驾驶车辆101、102、103，使得无人驾驶车辆101、102、103在遇到类似的突发事件时，能够通过端到端模型及时输出对应的控制指令，以规避突发事件。实际中，突发事件可能从不同的位置，或以不同的形式出现。依次，还需要采集端到端模型在无人驾驶车辆上应用时传感器采集的数据，即反馈数据。通过反馈数据对端到端模型进行修正，以提高端到端模型输出控制指令的准确性和鲁棒性。
在本实施例的一些可选的实现方式中，所述修正样本的采集方法为：
第一步，获取无人驾驶车辆的车辆数据。
在本实施例中，无人驾驶车辆的数据训练方法运行于其上的电子设备(例如图1所示的服务器105或无人驾驶车辆101、102、103上的无人驾驶车辆大脑)可以通过有线连接方式或者无线连接方式从无人驾驶车辆101、102、103获取无人驾驶车辆的车辆数据。无人驾驶车辆大脑可以是具有数据处理功能的车载电子终端设备。实践中，需要指出的是，上述无线连接方式可以包括但不限于3G/4G连接、WiFi连接、蓝牙连接、WiMAX连接、Zigbee连接、UWB(ultra wideband)连接、以及其他现在已知或将来开发的无线连接方式。
车辆数据可以包括无人驾驶车辆101、102、103上安装的各类传感器采集的数据，例如可以是距离传感器采集的无人驾驶车辆101、102、103与周围物体之间的距离数据；角度传感器采集的轮胎偏转方 向的角度数据；速度传感器采集的无人驾驶车辆101、102、103的行驶速度数据；温度传感器采集的发动机的工作温度数据等。车辆数据还可以包括无人驾驶车辆101、102、103的行驶状态数据，例如可以是无人驾驶车辆101、102、103当前处于匀速行驶状态、直线行驶状态，或车门处于开启状态等。
第二步，基于上述车辆数据，判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，之后从人工驾驶模式切换到无人驾驶模式。
由上述描述可知，当无人驾驶车辆101、102、103的无人驾驶与实际的路况之间的偏差达到一定程度时，无人驾驶车辆101、102、103可能无法正常执行车辆控制指令，从而容易出现不安全行驶的可能。此时就需要通过人工介入，使无人驾驶车辆101、102、103从无人驾驶模式切换到人工驾驶模式。当通过人工驾驶模式使无人驾驶车辆101、102、103的对应参数重新回到正常范围后，再从人工驾驶模式切换到无人驾驶模式。在这个过程中，可以将无人驾驶车辆101、102、103无法正常执行车辆控制指令的情况认为是干扰，对应的，人工介入后的人工驾驶模式过程可以认为是用来消除干扰的过程。当通过人工驾驶模式消除掉干扰后，无人驾驶车辆101、102、103再切换到无人驾驶模式。无人驾驶车辆101、102、103可以通过各种传感器记录上述的过程，得到对应的车辆数据。
正常情况下，无人驾驶车辆101、102、103可以按照需要设定的线路自动行驶。当有人工介入，使得无人驾驶车辆101、102、103由无人驾驶模式切换为人工驾驶模式，或由人工驾驶模式切换为无人驾驶模式时，服务器105或无人驾驶车辆大脑可以从车辆数据中找到对应上述模式切换时的数据。本实施例需要判断无人驾驶车辆101、102、103是否是先从无人驾驶模式切换为人工驾驶模式，之后又从人工驾驶模式切换到无人驾驶模式。由上述描述可知，从无人驾驶模式切换为人工驾驶模式，之后又从人工驾驶模式切换到无人驾驶模式的过程可以认为是一个出现干扰并消除干扰的过程。采集这个过程的车辆数据可以有助于提高无人驾驶车辆101、102、103的行驶安全。
在本实施例的一些可选的实现方式中，上述车辆数据可以包括传感器数据；以及，上述判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式可以包括：判断上述无人驾驶车辆的方向盘上的压力传感器所采集的压力数据是否大于预设的压力阈值；和/或判断上述无人驾驶车辆的方向盘上的温度传感器所采集的温度数据是否大于预设的温度阈值；若是，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
人工驾驶模式下，需要驾驶员对方向盘进行控制，此时，必然对方向盘施加压力。如此，就可以通过设置在方向盘上的压力传感器来监测是否处于人工驾驶模式。同理，人工驾驶模式下，驾驶员的手需要接触方向盘。此时，通过设置在方向盘上的温度传感器也可以检测到方向盘上的部分区域温度与其他区域不同的情况，进而监测到无人驾驶车辆101、102、103处于人工驾驶模式。
在本实施例的一些可选的实现方式中，上述车辆数据可以包括预期驾驶数据和实际驾驶数据；以及，上述判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式可以包括：判断上述预期驾驶数据与上述实际驾驶数据之间的差值是否大于预设的阈值；若是，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
无人驾驶车辆101、102、103可以按照预期驾驶数据进行无人驾驶，并同时记录无人驾驶车辆101、102、103的实际驾驶数据。虽然可能受到外部干扰，但只要预期驾驶数据和实际驾驶数据之间的差值在一定的误差范围内，都可以认为无人驾驶车辆101、102、103处于正常的无人驾驶。反之，当预期驾驶数据和实际驾驶数据之间的差值超过预设的阈值，就可以认为无人驾驶车辆101、102、103的行驶状态从无人驾驶模式切换到了人工驾驶模式。
上述描述了如何判断无人驾驶车辆101、102、103从无人驾驶模式切换到人工驾驶模式。类似的，可以通过上述的方法判断无人驾驶车辆101、102、103是否从人工驾驶模式切换到了无人驾驶模式，此处不再一一赘述。
第三步，响应于确定上述无人驾驶车辆从无人驾驶模式切换到人 工驾驶模式，确定切换到人工驾驶模式的第一切换时刻，响应于确定上述无人驾驶车辆从人工驾驶模式切换到无人驾驶模式，确定切换到无人驾驶模式的第二切换时刻。
当判断出无人驾驶车辆101、102、103在无人驾驶模式和人工驾驶模式之间切换后，就能够从车辆数据上找到对应的切换时刻。本实施例中，将无人驾驶车辆101、102、103从无人驾驶模式切换到人工驾驶模式的时刻确定为第一切换时刻；将无人驾驶车辆101、102、103从人工驾驶模式切换到无人驾驶模式的时刻确定为第二切换时刻。
第四步，将上述第一切换时刻和第二切换时刻之间获取的车辆数据标记为修正样本。
第一切换时刻为无人驾驶车辆101、102、103从无人驾驶模式切换为人工驾驶模式的时刻；第二切换时刻为无人驾驶车辆101、102、103从人工驾驶模式切换为无人驾驶模式的时刻。在第一切换时刻和第二切换时刻这段时间里，无人驾驶车辆101、102、103处于人工驾驶模式。由上述描述可知，人工驾驶模式的过程是个抗干扰的过程。因此，可以将第一切换时刻和第二切换时刻之间获取的车辆数据标记为修正样本。
在本实施例的一些可选的实现方式中，本实施例方法还可以包括：将上述第一切换时刻之前预设时段内获取的车辆数据标记为负样本。
基于背景技术的描述，无人驾驶车辆101、102、103在无法正常处于无人驾驶模式时，需要从无人驾驶模式切换到人工驾驶模式。可以认为无人驾驶车辆101、102、103在切换到人工驾驶模式前的一段时间里，无人驾驶车辆101、102、103受到了干扰(或遇到了突发事件)。因此，可以将第一切换时刻之前预设时段内获取的车辆数据标记为负样本。本实施例的负样本用来表征无人驾驶车辆101、102、103受到的干扰。其中，负样本用于表征无人驾驶车辆遇到的干扰，预设时段的具体取值根据实际需要而定。
在本实施例的一些可选的实现方式中，本实施例方法还可以包括：利用上述修正样本和负样本训练预设的无人驾驶车辆大脑模型，其中，上述无人驾驶车辆大脑模型用于基于车辆数据预测车辆控制指令。
上述可以认为是出现了干扰(负样本)，然后通过人工驾驶模式的控制(修正样本)消除了干扰的过程。因此，可以将负样本和修正样本作为训练样本来训练无人驾驶车辆大脑模型，使得无人驾驶车辆大脑模型能够从修正样本中学习到，在出现负样本的情况下，如何发出车辆控制指令来消除负样本造成的干扰。
继续参见图3，图3是根据本实施例的无人驾驶车辆的数据训练方法的应用场景的一个示意图。在图3的应用场景中，服务器105获取无人驾驶车辆的传感器数据和修正样本，之后，通过传感器数据和修正样本构建端到端模型，以使得无人驾驶车辆在遇到类似的干扰时能够及时输出正确的控制指令。
本申请的上述实施例提供的方法提高了无人驾驶车辆的行驶安全性。
进一步参考图4，作为对上述各图所示方法的实现，本申请提供了一种无人驾驶车辆的数据训练装置的一个实施例，该装置实施例与图2所示的方法实施例相对应，该装置具体可以应用于各种电子设备中。
如图4所示，本实施例的无人驾驶车辆的数据训练装置400可以包括：样本获取单元401和数据训练单元402。其中，样本获取单元401，用于获取无人驾驶车辆的传感器数据和修正样本，其中，所述修正样本用于表征无人驾驶车辆行驶过程中遇到干扰时，驾驶员的驾驶行为数据；数据训练单元402，用于通过所述传感器数据和修正样本构建端到端模型，所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。
在本实施例的一些可选的实现方式中，所述装置可以包括：修正单元(图中未示出)，用于将所述端到端模型推送给无人驾驶车辆，并通过测得的反馈数据对所述端到端模型进行修正。
在本实施例的一些可选的实现方式中，所述装置可以包括：修正样本采集单元(图中未示出)，用于采集修正样本，所述修正样本采集单元可以包括：车辆数据获取子单元(图中未示出)、驾驶模式判断子单元(图中未示出)、切换时刻确定子单元(图中未示出)和第一标记 子单元(图中未示出)。其中，车辆数据获取子单元用于获取无人驾驶车辆的车辆数据；驾驶模式判断子单元用于基于上述车辆数据，判断上述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，之后从人工驾驶模式切换到无人驾驶模式；切换时刻确定子单元用于响应于确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式，确定切换到人工驾驶模式的第一切换时刻，响应于确定上述无人驾驶车辆从人工驾驶模式切换到无人驾驶模式，确定切换到无人驾驶模式的第二切换时刻；第一标记子单元用于将上述第一切换时刻和第二切换时刻之间获取的车辆数据标记为修正样本。
在本实施例的一些可选的实现方式中，上述车辆数据可以包括传感器数据；以及上述驾驶模式判断子单元可以包括：第一判断模块(图中未示出)、第二判断模块(图中未示出)和第一确定模块(图中未示出)。其中，第一判断模块用于判断上述无人驾驶车辆的方向盘上的压力传感器所采集的压力数据是否大于预设的压力阈值；和/或，第二判断模块用于判断上述无人驾驶车辆的方向盘上的温度传感器所采集的温度数据是否大于预设的温度阈值；第一确定模块用于若上述压力数据大于预设的压力阈值和/或上述温度数据大于预设的温度阈值，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
在本实施例的一些可选的实现方式中，上述车辆数据可以包括预期驾驶数据和实际驾驶数据；以及上述驾驶模式判断子单元可以包括：第三判断模块(图中未示出)和第二确定模块(图中未示出)。其中，第三判断模块用于判断上述预期驾驶数据与上述实际驾驶数据之间的差值是否大于预设的阈值；第二确定模块用于若上述差值大于预设的阈值，则确定上述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
在本实施例的一些可选的实现方式中，无人驾驶车辆的数据训练装置400还可以包括：第二标记单元(图中未示出)，用于将上述第一切换时刻之前预设时段内获取的车辆数据标记为负样本，所述负样本用于表征无人驾驶车辆遇到的干扰。
本实施例还提供了一种服务器，包括：一个或多个处理器；存储器，用于存储一个或多个程序，当上述一个或多个程序被上述一个或 多个处理器执行时，使得上述一个或多个处理器执行上述的无人驾驶车辆的数据训练方法。
本实施例还提供了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现上述的无人驾驶车辆的数据训练方法。
下面参考图5，其示出了适于用来实现本申请实施例的服务器的计算机系统500的结构示意图。图5示出的服务器仅仅是一个示例，不应对本申请实施例的功能和使用范围带来任何限制。
如图5所示，计算机系统500包括中央处理单元(CPU)501，其可以根据存储在只读存储器(ROM)502中的程序或者从存储部分508加载到随机访问存储器(RAM)503中的程序而执行各种适当的动作和处理。在RAM 503中，还存储有系统500操作所需的各种程序和数据。CPU 501、ROM 502以及RAM 503通过总线504彼此相连。输入/输出(I/O)接口505也连接至总线504。
以下部件连接至I/O接口505：包括键盘、鼠标等的输入部分506；包括诸如阴极射线管(CRT)、液晶显示器(LCD)等以及扬声器等的输出部分507；包括硬盘等的存储部分508；以及包括诸如LAN卡、调制解调器等的网络接口卡的通信部分509。通信部分509经由诸如因特网的网络执行通信处理。驱动器510也根据需要连接至I/O接口505。可拆卸介质511，诸如磁盘、光盘、磁光盘、半导体存储器等等，根据需要安装在驱动器510上，以便于从其上读出的计算机程序根据需要被安装入存储部分508。
特别地，根据本公开的实施例，上文参考流程图描述的过程可以被实现为计算机软件程序。例如，本公开的实施例包括一种计算机程序产品，其包括承载在计算机可读介质上的计算机程序，该计算机程序包含用于执行流程图所示的方法的程序代码。在这样的实施例中，该计算机程序可以通过通信部分509从网络上被下载和安装，和/或从可拆卸介质511被安装。在该计算机程序被中央处理单元(CPU)501执行时，执行本申请的方法中限定的上述功能。
需要说明的是，本申请上述的计算机可读介质可以是计算机可读 信号介质或者计算机可读存储介质或者是上述两者的任意组合。计算机可读存储介质例如可以是——但不限于——电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。计算机可读存储介质的更具体的例子可以包括但不限于：具有一个或多个导线的电连接、便携式计算机磁盘、硬盘、随机访问存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑磁盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。在本申请中，计算机可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。而在本申请中，计算机可读的信号介质可以包括在基带中或者作为载波一部分传播的数据信号，其中承载了计算机可读的程序代码。这种传播的数据信号可以采用多种形式，包括但不限于电磁信号、光信号或上述的任意合适的组合。计算机可读的信号介质还可以是计算机可读存储介质以外的任何计算机可读介质，该计算机可读介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。计算机可读介质上包含的程序代码可以用任何适当的介质传输，包括但不限于：无线、电线、光缆、RF等等，或者上述的任意合适的组合。
附图中的流程图和框图，图示了按照本申请各种实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段、或代码的一部分，该模块、程序段、或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。也应当注意，在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个接连地表示的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或操作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
描述于本申请实施例中所涉及到的单元可以通过软件的方式实现， 也可以通过硬件的方式来实现。所描述的单元也可以设置在处理器中，例如，可以描述为：一种处理器包括样本获取单元和数据训练单元。其中，这些单元的名称在某种情况下并不构成对该单元本身的限定，例如，数据训练单元还可以被描述为“用于训练端到端模型的单元”。
作为另一方面，本申请还提供了一种计算机可读介质，该计算机可读介质可以是上述实施例中描述的装置中所包含的；也可以是单独存在，而未装配入该装置中。上述计算机可读介质承载有一个或者多个程序，当上述一个或者多个程序被该装置执行时，使得该装置：获取无人驾驶车辆的传感器数据和修正样本，其中，所述修正样本用于表征无人驾驶车辆行驶过程中遇到干扰时，驾驶员的驾驶行为数据；通过所述传感器数据和修正样本构建端到端模型，所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。
以上描述仅为本申请的较佳实施例以及对所运用技术原理的说明。本领域技术人员应当理解，本申请中所涉及的发明范围，并不限于上述技术特征的特定组合而成的技术方案，同时也应涵盖在不脱离上述发明构思的情况下，由上述技术特征或其等同特征进行任意组合而形成的其它技术方案。例如上述特征与本申请中公开的(但不限于)具有类似功能的技术特征进行互相替换而形成的技术方案。

Claims (14)

  1. 一种无人驾驶车辆的数据训练方法，其特征在于，所述方法包括：
    获取无人驾驶车辆的传感器数据和修正样本，其中，所述修正样本用于表征无人驾驶车辆行驶过程中遇到干扰时，驾驶员的驾驶行为数据；
    通过所述传感器数据和修正样本构建端到端模型，所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。
  2. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    将所述端到端模型推送给无人驾驶车辆，并通过测得的反馈数据对所述端到端模型进行修正。
  3. 根据权利要求1所述的方法，其特征在于，所述修正样本的采集方法为：
    获取无人驾驶车辆的车辆数据；
    基于所述车辆数据，判断所述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，之后从人工驾驶模式切换到无人驾驶模式；
    响应于确定所述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式，确定切换到人工驾驶模式的第一切换时刻，响应于确定所述无人驾驶车辆从人工驾驶模式切换到无人驾驶模式，确定切换到无人驾驶模式的第二切换时刻；
    将所述第一切换时刻和第二切换时刻之间获取的车辆数据标记为修正样本。
  4. 根据权利要求3所述的方法，其特征在于，所述车辆数据包括传感器数据；以及
    所述判断所述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶 模式，包括：
    判断所述无人驾驶车辆的方向盘上的压力传感器所采集的压力数据是否大于预设的压力阈值；和/或
    判断所述无人驾驶车辆的方向盘上的温度传感器所采集的温度数据是否大于预设的温度阈值；
    若是，则确定所述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
  5. 根据权利要求3所述的方法，其特征在于，所述车辆数据包括预期驾驶数据和实际驾驶数据；以及
    所述判断所述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，包括：
    判断所述预期驾驶数据与所述实际驾驶数据之间的差值是否大于预设的阈值；
    若是，则确定所述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
  6. 根据权利要求3所述的方法，其特征在于，所述方法还包括：
    将所述第一切换时刻之前预设时段内获取的车辆数据标记为负样本，所述负样本用于表征无人驾驶车辆遇到的干扰。
  7. 一种无人驾驶车辆的数据训练装置，其特征在于，所述装置包括：
    样本获取单元，用于获取无人驾驶车辆的传感器数据和修正样本，其中，所述修正样本用于表征无人驾驶车辆行驶过程中遇到干扰时，驾驶员的驾驶行为数据；
    数据训练单元，用于通过所述传感器数据和修正样本构建端到端模型，所述端到端模型用于通过传感器数据和修正样本输出对应驾驶员的驾驶行为的控制指令。
  8. 根据权利要求7所述的装置，其特征在于，所述装置包括：
    修正单元，用于将所述端到端模型推送给无人驾驶车辆，并通过测得的反馈数据对所述端到端模型进行修正。
  9. 根据权利要求7所述的装置，其特征在于，所述装置还包括：
    修正样本采集单元，用于采集修正样本，所述修正样本采集单元包括：
    车辆数据获取子单元，用于获取无人驾驶车辆的车辆数据；
    驾驶模式判断子单元，用于基于所述车辆数据，判断所述无人驾驶车辆是否从无人驾驶模式切换到人工驾驶模式，之后从人工驾驶模式切换到无人驾驶模式；
    切换时刻确定子单元，用于响应于确定所述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式，确定切换到人工驾驶模式的第一切换时刻，响应于确定所述无人驾驶车辆从人工驾驶模式切换到无人驾驶模式，确定切换到无人驾驶模式的第二切换时刻；
    第一标记子单元，用于将所述第一切换时刻和第二切换时刻之间获取的车辆数据标记为修正样本。
  10. 根据权利要求9所述的装置，其特征在于，所述车辆数据包括传感器数据；以及
    所述驾驶模式判断子单元包括：
    第一判断模块，用于判断所述无人驾驶车辆的方向盘上的压力传感器所采集的压力数据是否大于预设的压力阈值；和/或
    第二判断模块，用于判断所述无人驾驶车辆的方向盘上的温度传感器所采集的温度数据是否大于预设的温度阈值；
    第一确定模块，用于若所述压力数据大于预设的压力阈值和/或所述温度数据大于预设的温度阈值，则确定所述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
  11. 根据权利要求9所述的装置，其特征在于，所述车辆数据包 括预期驾驶数据和实际驾驶数据；以及
    所述驾驶模式判断子单元包括：
    第三判断模块，用于判断所述预期驾驶数据与所述实际驾驶数据之间的差值是否大于预设的阈值；
    第二确定模块，用于若所述差值大于预设的阈值，则确定所述无人驾驶车辆从无人驾驶模式切换到人工驾驶模式。
  12. 根据权利要求9所述的装置，其特征在于，所述装置还包括：
    第二标记单元，用于将所述第一切换时刻之前预设时段内获取的车辆数据标记为负样本，所述负样本用于表征无人驾驶车辆遇到的干扰。
  13. 一种服务器，包括：
    一个或多个处理器；
    存储装置，用于存储一个或多个程序，
    当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如权利要求1-6中任一所述的方法。
  14. 一种计算机可读存储介质，其上存储有计算机程序，其特征在于，该程序被处理器执行时实现如权利要求1-6中任一所述的方法。
PCT/CN2018/099173 2017-09-05 2018-08-07 无人驾驶车辆的数据训练方法和装置 Ceased WO2019047657A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP18853701.3A EP3582056B1 (en) 2017-09-05 2018-08-07 Data training method and device for autonomous vehicle
US16/566,826 US20200004249A1 (en) 2017-09-05 2019-09-10 Data training method and apparatus for autonomous vehicle

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710790859.0A CN107491073B (zh) 2017-09-05 2017-09-05 无人驾驶车辆的数据训练方法和装置
CN201710790859.0 2017-09-05

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/566,826 Continuation US20200004249A1 (en) 2017-09-05 2019-09-10 Data training method and apparatus for autonomous vehicle

Publications (1)

Publication Number Publication Date
WO2019047657A1 true WO2019047657A1 (zh) 2019-03-14

Family

ID=60651289

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/099173 Ceased WO2019047657A1 (zh) 2017-09-05 2018-08-07 无人驾驶车辆的数据训练方法和装置

Country Status (4)

Country Link
US (1) US20200004249A1 (zh)
EP (1) EP3582056B1 (zh)
CN (1) CN107491073B (zh)
WO (1) WO2019047657A1 (zh)

Families Citing this family (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107491073B (zh) * 2017-09-05 2021-04-02 百度在线网络技术（北京）有限公司 无人驾驶车辆的数据训练方法和装置
CN107571864B (zh) * 2017-09-05 2019-11-05 百度在线网络技术（北京）有限公司 无人驾驶车辆的数据采集方法和装置
US10656644B2 (en) * 2017-09-07 2020-05-19 Tusimple, Inc. System and method for using human driving patterns to manage speed control for autonomous vehicles
US10866588B2 (en) * 2017-10-16 2020-12-15 Toyota Research Institute, Inc. System and method for leveraging end-to-end driving models for improving driving task modules
CN109131356B (zh) * 2018-09-07 2020-12-08 泉州台商投资区五逸季科技有限公司 人机混合增强智能驾驶系统及电动汽车
CN110968088B (zh) * 2018-09-30 2023-09-12 百度（美国）有限责任公司 车辆控制参数的确定方法、装置、车载控制器和无人车
CN110967991B (zh) * 2018-09-30 2023-05-26 百度（美国）有限责任公司 车辆控制参数的确定方法、装置、车载控制器和无人车
CN109597412A (zh) * 2018-12-06 2019-04-09 江苏萝卜交通科技有限公司 一种无人驾驶系统及其控制方法
CN109858369A (zh) 2018-12-29 2019-06-07 百度在线网络技术（北京）有限公司 自动驾驶方法和装置
CN110019688A (zh) * 2019-01-23 2019-07-16 艾肯特公司 对机器人进行训练的方法
CN109857118B (zh) * 2019-03-12 2022-08-16 百度在线网络技术（北京）有限公司 无人车规划驾驶策略的方法、装置、设备及存储介质
CN110069064B (zh) * 2019-03-19 2021-01-29 驭势科技（北京）有限公司 一种自动驾驶系统升级的方法、自动驾驶系统及车载设备
CN110478911A (zh) * 2019-08-13 2019-11-22 苏州钛智智能科技有限公司 基于机器学习的智能游戏车无人驾驶方法及智能车、设备
CN110481630B (zh) * 2019-08-21 2021-07-20 同济大学 一种基于电子皮肤技术的方向盘及交互方法
GB201912145D0 (en) * 2019-08-23 2019-10-09 Five Ai Ltd Performance testing for robotic systems
US11711906B2 (en) * 2020-02-04 2023-07-25 Tusimple, Inc. Temperature management system for autonomous vehicles
CN112269385B (zh) * 2020-10-23 2021-09-07 北京理工大学 云端无人车动力学控制系统和方法
CN112836395B (zh) * 2021-03-10 2024-07-02 北京车和家信息技术有限公司 一种车辆行驶数据模拟方法、装置、电子设备及存储介质
CN113641170B (zh) * 2021-08-31 2023-02-03 北京三快在线科技有限公司 一种智能化检测无人驾驶接管问题的方法及装置
CN115230718A (zh) * 2021-11-23 2022-10-25 上海仙途智能科技有限公司 无人驾驶清扫车的控制方法、系统和可读存储介质
CN114030484B (zh) * 2021-12-21 2023-06-02 华人运通(江苏)技术有限公司 自动驾驶功能控制方法、装置、设备和存储介质
CN114489714A (zh) * 2021-12-24 2022-05-13 北京百度网讯科技有限公司 一种车载数据处理方法、装置、电子设备及存储介质
CN115352468B (zh) * 2022-08-31 2025-10-14 际络科技(上海)有限公司 多驾驶状态的自动驾驶数据获取方法、设备及介质
CN115798190A (zh) * 2022-10-26 2023-03-14 上海仙途智能科技有限公司 车辆接管状态确定方法及装置、终端设备、存储介质
CN116956044A (zh) * 2023-08-01 2023-10-27 北京大示示粒科技有限公司 自动驾驶车辆及其性能评估方法、评估系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7512487B1 (en) * 2006-11-02 2009-03-31 Google Inc. Adaptive and personalized navigation system
CN104108395A (zh) * 2013-04-16 2014-10-22 福特全球技术公司 用于修改机动车辆的驾驶辅助系统的配置的方法和装置
CN105711591A (zh) * 2016-04-26 2016-06-29 百度在线网络技术（北京）有限公司 无人驾驶车辆、无人驾驶车辆的控制方法和装置
CN106503393A (zh) * 2016-11-15 2017-03-15 浙江大学 一种利用仿真生成样本实现无人车自主行进的方法
CN107491073A (zh) * 2017-09-05 2017-12-19 百度在线网络技术（北京）有限公司 无人驾驶车辆的数据训练方法和装置
CN107564363A (zh) * 2017-09-05 2018-01-09 百度在线网络技术（北京）有限公司 一种用于驾驶模式切换的方法和装置

Family Cites Families (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6173218B1 (en) * 1997-10-15 2001-01-09 The Boeing Company Neurocomputing control distribution system
JP3870911B2 (ja) * 2003-02-10 2007-01-24 日産自動車株式会社 車線逸脱防止装置
JP5272448B2 (ja) * 2008-03-04 2013-08-28 日産自動車株式会社 車両用運転支援装置及び車両用運転支援方法
CN102426019B (zh) * 2011-08-25 2014-07-02 航天恒星科技有限公司 一种无人机景象匹配辅助导航方法及系统
WO2015159341A1 (ja) * 2014-04-14 2015-10-22 三菱電機株式会社 運転支援装置および運転支援方法
US9555807B2 (en) * 2015-05-01 2017-01-31 Delphi Technologies, Inc. Automated vehicle parameter modification based on operator override
JP6376059B2 (ja) * 2015-07-06 2018-08-22 トヨタ自動車株式会社 自動運転車両の制御装置
JP6773040B2 (ja) * 2015-09-30 2020-10-21 ソニー株式会社 情報処理システム、および情報処理システムの情報処理方法、情報処理装置、並びにプログラム
US9956956B2 (en) * 2016-01-11 2018-05-01 Denso Corporation Adaptive driving system
US10579070B1 (en) * 2016-01-22 2020-03-03 State Farm Mutual Automobile Insurance Company Method and system for repairing a malfunctioning autonomous vehicle
CN105892471B (zh) * 2016-07-01 2019-01-29 北京智行者科技有限公司 汽车自动驾驶方法和装置
CN106080606B (zh) * 2016-07-08 2019-01-01 百度在线网络技术（北京）有限公司 用于控制无人驾驶车辆的方法和装置
CN106154834B (zh) * 2016-07-20 2019-10-29 百度在线网络技术（北京）有限公司 用于控制无人驾驶车辆的方法和装置
DE102016224291A1 (de) * 2016-12-06 2018-06-07 Trw Automotive Gmbh Verfahren zur rechnergestützten Adaption eines vorgegebenen teilautomatisierten Fahrsystems eines Kraftfahrzeugs
US10133270B2 (en) * 2017-03-28 2018-11-20 Toyota Research Institute, Inc. Electronic control units, vehicles, and methods for switching vehicle control from an autonomous driving mode
EP3452352A4 (en) * 2017-06-30 2019-07-10 Beijing Didi Infinity Technology and Development Co., Ltd. SYSTEMS AND METHOD FOR SWITCHING THE DRIVING MODE OF A VEHICLE
US10289114B2 (en) * 2017-08-08 2019-05-14 The Boeing Company Safety controls for network connected autonomous vehicle

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7512487B1 (en) * 2006-11-02 2009-03-31 Google Inc. Adaptive and personalized navigation system
CN104108395A (zh) * 2013-04-16 2014-10-22 福特全球技术公司 用于修改机动车辆的驾驶辅助系统的配置的方法和装置
CN105711591A (zh) * 2016-04-26 2016-06-29 百度在线网络技术（北京）有限公司 无人驾驶车辆、无人驾驶车辆的控制方法和装置
CN106503393A (zh) * 2016-11-15 2017-03-15 浙江大学 一种利用仿真生成样本实现无人车自主行进的方法
CN107491073A (zh) * 2017-09-05 2017-12-19 百度在线网络技术（北京）有限公司 无人驾驶车辆的数据训练方法和装置
CN107564363A (zh) * 2017-09-05 2018-01-09 百度在线网络技术（北京）有限公司 一种用于驾驶模式切换的方法和装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
HAO YU .: "Baidu Driving Dataset and End-to-End Reactive Control Model", 2017 IEEE INTELLIGENT VEHICLES SYMPOSIUM, 14 June 2017 (2017-06-14), XP033133730 *

Also Published As

Publication number Publication date
CN107491073A (zh) 2017-12-19
US20200004249A1 (en) 2020-01-02
CN107491073B (zh) 2021-04-02
EP3582056A1 (en) 2019-12-18
EP3582056B1 (en) 2025-05-14
EP3582056A4 (en) 2020-11-04

Similar Documents

Publication Publication Date Title
WO2019047657A1 (zh) 无人驾驶车辆的数据训练方法和装置
JP7174063B2 (ja) 無人運転車用の障害物回避方法及び装置
US10311652B2 (en) Method and device for modifying the configuration of a driving assistance system of a motor vehicle
WO2019047650A1 (zh) 无人驾驶车辆的数据采集方法和装置
US11842638B2 (en) Parallel remote control driving system for intelligent network vehicle
CN109421630B (zh) 用于监测自主车辆的健康的控制器架构
US20220139092A1 (en) Model generating apparatus, method and program, and prediction apparatus
WO2019047646A1 (zh) 车辆避障方法和装置
WO2019047651A1 (zh) 驾驶行为预测方法和装置、无人车
WO2019047656A1 (zh) 用于控制无人驾驶车辆的方法和装置
US11628861B2 (en) Method and apparatus for controlling vehicle, device and storage medium
CN115019267A (zh) 车辆场景数据的采集方法、装置、存储介质及电子设备
US20190064813A1 (en) Systems and Methods of Controlling an Autonomous Vehicle Using an Enhanced Trajectory Following Configuration
MX2015002632A (es) Monitoreo del operador del vehiculo y ajustes de las operaciones.
EP3570214B1 (en) Automobile image processing method and apparatus, and readable storage medium
CN110696826A (zh) 用于控制车辆的方法和装置
US11654938B1 (en) Methods and apparatus for disengaging an autonomous mode based on lateral error of an autonomous vehicle
CN109094574B (zh) 一种基于台架的无人车驾驶状态测量控制系统
WO2022245916A1 (en) Device health code broadcasting on mixed vehicle communication networks
Rojas et al. Performance evaluation of an autonomous vehicle using resilience engineering
US20180074756A1 (en) Method and apparatus for acquiring data in a robot operating system
US12024100B2 (en) Device-level fault detection
CN110588666B (zh) 用于控制车辆行驶的方法及装置
CN114968189A (zh) 用于自动驾驶系统的感知系统开发的平台
CN115440031B (zh) 一种考虑路侧感知性能的车路协同混合车辆队列控制方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18853701

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2018853701

Country of ref document: EP

Effective date: 20190912

NENP Non-entry into the national phase

Ref country code: DE

WWG Wipo information: grant in national office

Ref document number: 2018853701

Country of ref document: EP