SE1450739A1

SE1450739A1 - System for controlling the deployment of an external safety device

Info

Publication number: SE1450739A1
Application number: SE1450739A
Authority: SE
Inventors: Laurent Caclard
Original assignee: Autoliv Dev
Priority date: 2014-06-16
Filing date: 2014-06-16
Publication date: 2015-05-28

Abstract

The present invention relates to a vehicle safety system comprising an external safety system for protecting a pedestrian, and a device for activating the external safety system (13, 14) the device being connected to a pre-crash sensor system (10, 11, 15) and an in-crash sensor system (16). The device is arranged to perform an object classification of a first signal (Al) based on the output from object and thermal sensors. The device is further arranged to determine a threshold (Tl, T2) for the in-crash sensor system (16) as a function of the object classification; where the device is arranged to set a first threshold (Tl) as a function of a first signal (Al) indicating no object detected or an object classified as a pedestrian; and to set a second threshold (T2), higher than the first threshold (Tl), as a function of a first signal (Al) indicating an object classified as a non-pedestrian; and that the device is arranged to compare a second signal (A2) from the in-crash sensor system (16) with the threshold (Tl, T2), whereby the external safety system (13, 14) is activated if the set threshold (Tl, T2) is exceeded.(Fig. 1)

Description

44 ABSTRACT The present invention relates to a vehicle safety system comprising an external safety system for protecting a pedestrian, and a device for activating the external safety system (13, 14) the device being connected to a pre-crash sensor system (10, 11, 15) and an in-crash sensor system (16). The device is arranged to perform an object classification of a first signal (Al) based on the output from object and thermal sensors. The device is further arranged to determine a threshold (T1, T2) for the in-crash sensor system (16) as a function of the object classification; where the device is arranged to set a first threshold (T1) as a function of a first signal (Al) indicating no object detected or an object classified as a pedestrian; and to set a second threshold (T2), higher than the first threshold (T1), as a function of a first signal (Al) indicating an object classified as a non-pedestrian; and that the device is arranged to compare a second signal (A2) from the in-crash sensor system (16) with the threshold (T1, 12), whereby the external safety system (13, 14) is activated if the set threshold (T1, T2) is exceeded.

(Fig. 1) 1 TITLE System for controlling the deployment of an external safety device DESCRIPTION OF THE INVENTION The present invention relates to a system for sensing a motor vehicle impact and adjusting the deployment of an external safety device. 10 Today, many vehicles are constructed to comprise at least one exterior airbag, intended to be inflated in the case of a collision with moving objects, such as pedestrians, cyclists or animals and to alleviate the collision force that the moving object is subjected to.

In order to further minimize the damages afflicted to the moving object in the event of such an impact, it has been proposed to use some kind of hood lifting arrangements. These arrangements are generally constructed so that the rear part of the hood, i.e. the part closest to the windscreen, is lifted in the event of a collision with a moving object. Arrangements of this type are disclosed in for example W02007/067121 and EP2256007.

For any external airbag system to operate properly, a robust sensing system is necessary. Unlike crash sensors which trigger deployment while the vehicle is crushing and decelerating, the sensing system for an external airbag may anticipate an impact before it has occurred. This critical "Time Before Collision" is related to the time to deploy the actuator (e.g. 30-200 ms) and the clearance distance in front of the vehicle (e.g. 100-800 mm). Inadvertent deployment is 2 not only costly but may temporarily disable the vehicle. Moreover, since the deployment of an airbag is achieved through a release of energy, deployment at an inappropriate time may result in undesirable effects. This invention is related to a sensing system for an external airbag safety system which addresses these design concerns.

Radar and ultrasonic detection systems have been studied and employed for motor vehicles for many years. Radar systems for motor vehicles operate in that a radio frequency signal, typically in the microwave region, is emitted from an antenna on the vehicle and the reflected-back signal is analysed to reveal information about the reflecting target. Such systems have been considered for use in active braking systems for motor vehicles, as well as obstacle detection systems for vehicle drivers. Radar sensing systems also have applicability in deploying external airbags. Radar sensors provide a number of valuable inputs, including the ability to detect the range to the closest object with a high degree of accuracy (e.g. cm). They can also provide an output enabling measurement of closing velocity to a target with high accuracy. The radar cross section of the target and the characteristics of the return signal may also be used as a means of characterizing and identifying moving objects.

An alternative means for detecting and identifying moving objects is a vision system comprising one or more video cameras used for detecting objects. A vision system for a vehicle identifies and classifies objects located proximate a vehicle. The system comprises a sensor array that produces imagery that is processed to generate depth maps of the scene proximate a vehicle. The depth maps are processed and compared to pre-rendered templates of target objects that could appear 3 proximate the vehicle. A target list is produced by matching the pre-rendered templates to the depth map imagery. The system processes the target list to produce target size and classification estimates. The target is then tracked as it moves near a vehicle and the target position, classification and velocity are determined. A further alternative object detection system for a vehicle comprises an infrared camera for gathering an image of at least a part of the surroundings of the vehicle; and a processor for applying an algorithm to at least a part of the image gathered by the camera, the algorithm identifying non-relevant hot or warm objects detected by the camera and reducing the brightness and/or distinctiveness of the non-relevant objects in the image.

A classification system is used for classifying objects in the vicinity of a vehicle. The system typically comprises a video/infrared camera for gathering camera data, a reflected radiation system for gathering reflected radiation data, and a classifier. Raw data from the video/infrared camera and the reflected radiation system can be combined and analysed by the classifier, the classifier being configured to provide an output relating to the type of an object that appears in data gathered by both the camera and the reflected radiation system.

Existing object classifiers comprise computer programs which are operable to analyse data from a vehicle sensor, such as a camera or radar system. The classifier is trained with exposure to many different types of object in different circumstances, so that the program is able to make an accurate determination as to the type of a new kind of object that is detected. Known types of objects can be stored in a database for future reference. In the subsequent text, the term "pre- 4 crash sensor" will be used for vehicle sensors comprising a camera or a radar/ultrasonic system.

A vehicle is also provided with one or more contact sensors, 5 such as accelerometers or pressure sensors, for detecting an actual impact. In the subsequent text, the term 'in-crash sensor" will also be used for this type of contact sensors. Modern vehicles can be provided with 5-10 contact sensors across a front bumper or similar suitable portion of the 10 vehicle.

When it is determined by vehicle crash sensors that an impact is imminent, or that crash is occurring, one or more of these external safety systems may be deployed, or a safety system may be deployed in one of a plurality of possible modes, depending in part upon the type of the other object that is involved. If it appears that the vehicle is about to strike a pedestrian, then the external safety system in the form of an external air-bag and/or a bonnet lifter may be activated, but if the vehicle is about to strike an inanimate object such as a tree then there is no need for these protection systems to be deployed. Accurate classification of objects in the vicinity of a vehicle is therefore desirable for vehicle safety systems to be activated in the most appropriate manner.

Although information obtained from pre-crash sensors yield valuable data, exclusive reliance upon a pre-crash sensor signal for deploying an external airbag has certain negative consequences. As mentioned previously, deployment of the external airbag is a significant event and should only occur when needed in an impending impact situation. Pre-crash sensors are prone to "false-positive" (FP) indications. These are typically due to phenomena such as a ground reflection, projection of small objects, and software misinterpretation, which faults are also referred to as "fooling" and "ghosting". For example, a small metal object with a reflector type geometry can return as much energy as a small car and as such 5 can generate a collision signal in the radar even when the object is too small to damage the vehicle in a substantial way. Also, there may be "near miss" situations where a target is traveling fast enough to avoid collision, yet the pre-crash sensor would provide a triggering signal for the external 10 safety system.

The relative speed is a measure for the classification of objects. It can be used to establish, for example, whether the relevant object is at rest or whether it moves at a certain speed. Since pedestrians, for example, only have limited maximum speed, it is easy to distinguish pedestrians from vehicles. This may then be used in particular also to take into account the severity of an accident. Thus this makes it possible for the device according to the present invention to be able to make a triggering decision for the actuator system on the basis of speed. With the aid of speed information, the impact signal becomes easier to differentiate as to whether it involves a pedestrian, a cyclist or another object. Including the speed thus helps to prevent a false triggering of the actuator system. Overall this results in a more precise evaluation of the impact signal, that is, the signal from the contact sensor system. For example, a fast and light object may provide a similar impact signal as a slow and heavy object. This shows that the knowledge of the speed expands the decision space by a dimension such that a classification of the different objects is improved and thereby also the decision for triggering the actuator system. In the subsequent text, the term 'pedestrian" should be interpreted as including 6 any moving object for which the external safety system is suitable, i.e. pedestrians and cyclists, as well as animals.

It is therefore an object of the present invention to provide 5 an improved vehicle safety system to overcome or to mitigate the above problems.

The invention relates to a vehicle safety system comprising an external safety system for protecting a pedestrian. The vehicle safety system is also provided with a device for activating the external safety system, which device is connected to a pre-crash sensor system and an in-crash sensor system, of the types indicated above. The device is arranged to perform an object classification of a first signal from the pre-crash sensor system. First, the object classification determines whether an object is detected or not. Second, if an object is detected, the object classification determines whether the object is moving. In this case, the object classification can also determine if a non-moving object is a pedestrian or a fixed object. Third, if the object is moving, the object classification can determine whether the object is a pedestrian, a cyclist or similar.

The object detection is performed by the pre-crash sensor system which comprises at least one object sensor and at least one thermal sensor, arranged to detect the temperature, or heat signature, of an object. In this text, the invention will be described in the context of a pedestrian protection system. However, the system can also be applied to the detection of animals, in particular animals (mammals) over a predetermined size. This size can be dependent on the shape of the front of the vehicle, wherein animals large enough to be thrown onto 7 the bonnet and/or the windscreen are capable of causing injury to vehicle occupants in a collision.

The device is arranged to determine a threshold for the in- crash sensor system as a function of the object classification. The device is arranged to set a first threshold as a function of the first signal indicating no object detected or an object classified as a pedestrian. Also, the device is arranged to set a second threshold, higher than the first threshold, as a function of the first signal indicating an object classified as a non-pedestrian. Finally, the device is arranged to compare a second signal from the in-crash sensor system with the threshold. If the comparison indicates that at least one predetermined condition is fulfilled, the external safety system is activated if the set threshold is exceeded.

According to the invention, the device can be arranged to weight the first threshold as a function of a first signal based on the confidence level of the object and thermal sensors, respectively. For instance, if the object sensor detects an object with a high confidence level, then the output signal from the thermal sensor can be used to confirm the detection of an object classified as a pedestrian. If the object sensor detects an object with a low confidence level while the output signal from the thermal sensor has a high confidence level, then this indicates the detection of an object classified as a pedestrian. On the other hand, if the output signals from the object and thermal sensors have a low confidence level, then this indicates that an object classified as a non-pedestrian is detected. 8 Further, the device can be arranged to weight the first threshold as a function of a first signal based on a comparison between the detected object temperature and the ambient temperature. In order to differentiate a pedestrian, or a mammal, from the ambient background and from moving/inanimate objects (non-pedestrians), the thermal sensor is preferably, but not necessarily arranged to detect objects in a temperature window of approximately 30-42 °C. This temperature range covers commonly occurring core and surface temperatures for pedestrians and most mammals. In this way the output signal from the thermal sensor can be used as an additional means for filtering false positive and/or false negative indications from the one or more object sensors.

The discrimination capability of an infrared sensor system can be reduced under hot conditions or in tropical climates where background objects can have a temperature exceeding °C, e.g. in the daytime when the sun is shining. Under such conditions, the system will mainly rely on object sensors.

The system can also be arranged to fuse the output signals from the object and thermal sensors into a first signal. This fused signal forms the first signal from the pre-crash sensor system used for performing an object classification.

According to the invention the at least one thermal sensor is preferably an infra-red (IR) sensor. The at least one object sensor is a radar and/or a vision sensor. In this context the term "radar" is used for an arrangement detecting reflected pulses from an emitter on the vehicle, which pulses can have any suitable wavelength. The at least one vision sensor can be a suitable image detecting device, such as a camera. 9 According to the invention, the device is arranged to set a first threshold, lower than a nominal threshold, as a function of a first signal indicating an object classified as a pedestrian. In this context, the nominal threshold is a stored threshold that is used by the device when no input is available from the active sensor systems.

The device is arranged to set a second threshold, higher than the nominal threshold, as a function of a first signal 10 indicating an object classified as a non-pedestrian; Further, the device is arranged to interpolate the set threshold between the nominal threshold and a threshold limit value as a function of the probability of accurate 15 classification for the first signal.

The device is arranged compare a second signal from the in-crash sensor system with the set threshold, whereby the external safety system is activated if the set threshold is 20 exceeded.

In this context, the device can set a first threshold interpolated between the nominal threshold and a minimum value, if the classification indicates that it is probable that an object classified as a pedestrian has been detected. Similarly, the device can set a second threshold interpolated between the nominal threshold and a maximum value, if the classification indicates that it is probable that an object classified as a non-pedestrian has been detected. A more detailed description on the setting of the above thresholds dependent on the object classification can be found in the international application PCT/ SE2013/051035, which is hereby incorporated by reference.

According to a first alternative example, the device is arranged to set a first threshold equal to a minimum threshold limit value if the probability of accurate classification has a maximum value, as a function of the first signal indicating that an object classified as a pedestrian is detected.

According to a second alternative example, the device is arranged to set a first threshold higher than the minimum threshold limit value if the probability of accurate classification is between a minimum and a maximum value, as a function of the first signal indicating no object or that an object classified as a possible pedestrian is detected.

According to a third alternative example, the device is arranged to set a second threshold equal to a maximum threshold limit value if the probability of accurate classification has a maximum value, as a function of the first signal indicating that an object classified as a non- pedestrian is detected.

According to a fourth alternative example, the device is arranged to set a second threshold lower than the maximum threshold limit value if the probability of accurate classification between a minimum and a maximum value, as a function of the first signal (Al) indicating that an object classified as a possible non-pedestrian is detected.

According to a fifth alternative example the device is arranged to set a threshold equal to the nominal threshold, as a function of a first signal indicating that no input is available from the pre-crash sensor or pre-crash sensor failure. This situation can occur, for instance, for a sensor 11 which can only detect some types of objects but does not have a generic object classification, like mono-vision system In this case, the input from the in-crash sensors will be decisive for the triggering of the external safety system.

According to a further alternative example the device is arranged to set a threshold equal to a minimum threshold limit value, as a function of a first signal indicating no object detected. This situation can occur, for instance, for a sensor where any object impacting the car will normally be seen, such as a radar system. For such systems, a standing object in the path of the vehicle is very likely to be seen. However, because such sensors have a limited field of view a moving target passing in front of the vehicle may be hit before being detected by a sensor. In this case, it can be assumed that if the object was not seen, it is because it has moved out in front of the vehicle. Such an object is likely to be a pedestrian. Therefore a threshold equal to a minimum threshold limit value should be set.

For the vehicle safety system described above, the probability of accurate classification can be a function of the first signal and at least one additional detected vehicle related parameter. Examples of suitable vehicle related parameters are a detected brake actuation, such as the degree of brake actuation performed by the driver or an automatic system, and/or a detected steering wheel actuation, such as the rate of change of steering wheel angle.

The threshold set by the above vehicle safety system device can be a linear function of the probability of accurate classification between the maximum and minimum threshold limit 12 values. Further, the probability of accurate classification is a function of the sensors confidence level.

The sensor confidence level comprises generic information broadcast along with any signal from the sensor, and represents which confidence the signal provider has on the information to be accurate. For example, the vehicle speed provided on a CAN bus in the vehicle has a confidence level; if some of the ABS wheel speeds are not available from the respective speed sensors, the speed can still be estimated but less precisely. In this case the confidence level can be decrease to warn the other systems using this information that the input signal is less accurate than usual. Hence, the confidence level is an inherent property of the sensor. In the specific case of a pre-crash sensor, the confidence level can be based on environmental conditions (rain, snow, ambient light) or sensor diagnostics.

The device can further be arranged to determine a speed threshold relative to a nominal speed threshold for the in-crash sensor system as a function of the object classification. A vehicle safety system of this type can be operative in a predetermined vehicle speed range, for instance between 20 km/h and 50 km/h. In this case, the nominal speed threshold is the minimum speed at which the vehicle safety system is active, that is 20 km/h.

According to a first example the first signal indicates that no object is detected. In this case the device is arranged to set the speed threshold higher than the nominal speed threshold as a function of the first signal. The minimum speed at which the vehicle safety system is active is set to, for instance, 30 km/h to make the operative range 30-50 km/h. 13 According to a second example the first signal indicates that an object classified as a pedestrian is detected. In this case the device is arranged to set the speed threshold lower than the nominal speed threshold as a function of the first signal. The minimum speed at which the vehicle safety system is active is set to, for instance, 15 km/h to make the operative range 15-50 km/h.

According to a third example the first signal indicates that an object classified as a non-pedestrian is detected. In this case the device is arranged to set the speed threshold higher than the nominal speed threshold as a function of the first signal. The minimum speed at which the vehicle safety system is active is set to, for instance, 30 km/h to make the operative range 30-50 km/h.

The use of a speed threshold in this context is only briefly discussed here. A detailed description of a device using speed thresholds that is dependent on the object classification can be found in the international application PCT/ SE2013/051033, which is hereby incorporated by reference.

According to the invention, the external safety system can further comprise two or more zones, which zones are provided with a pre-crash sensor system and separately controlled external safety systems for each zone. Examples of such zones can comprise a first and a second zone covering the right and left side of the vehicle, respectively, or a right, left and centre zone covering the front of the vehicle. Preferably, but not necessarily, a single pre-crash sensor system can be configured to cover all zones using a suitable camera or a radar/ultrasonic system as described above. 14 Alternatively, adjacent zones can be provided with individual pre-crash sensor systems arranged to provide continuous cover, in order to avoid any dead angles.

After performing an object classification, which has been described above, the device is arranged to determine a threshold for the in-crash sensor system as a function of the object classification. The device is arranged to determine an individual threshold for each zone, as a function of a first signal from the pre-crash sensor system. The pre-crash sensor system is arranged to monitor all zones and will issue an individual first signal for each zone. The external safety system in each zone can be activated individually if the set threshold for a particular zone is exceeded. The use of multiple zones allows for multiple target tracking, that is, separate targets can be tracked in each zone.

For instance, if the vehicle is travelling along a continuous obstacle, such as a snow bank, the pre-crash sensor system can generate a first signal for the zone adjacent the obstacle that is used for performing an object classification. When it is determined that a fixed object is detected, a higher threshold can be set for this zone. This threshold is set independently of the thresholds set for any other zone. As indicated above, the device is arranged to determine a threshold for the in-crash sensor system as a function of the object classification. For example, the device can be arranged to set a first threshold as a function of the first signal indicating no object detected or an object classified as a pedestrian for the zone remote from the continuous obstacle. At the same time, the device can be arranged to set a second threshold, higher than the first threshold, as a function of the first signal indicating an object classified as a non-pedestrian, e.g. a snow bank, for the zone adjacent the continuous obstacle. This arrangement reduces the risk of false triggering of the external safety system in one zone, caused by a detected non-pedestrian objects, while allowing triggering of the external safety system in another zone, caused by a detected pedestrian.

A vehicle safety system according to the invention preferably comprises a pre-crash sensor system that has a field of view of less than 180°. When using multiple zones, the sum of all individual angles of field of view for these zones is less than 180°. For instance, for two zones, covering the right and left side of the vehicle respectively, the field of view for each zone is less than 90°. For three zones, covering the right and left side and a central sector of the vehicle respectively, the field of view for each zone is less than 60°. The left/right zones can have the same angle of view, but do not necessarily have the same angle of view as the central zone. For an external safety system for protecting a pedestrian, the field of view is inherently measured relative to the main direction of travel of the vehicle (see Fig.1).

The external safety system preferably comprises components arranged to be activated sequentially. A system comprising multiple components can be provided with at least a pyrotechnic hood lifter and at least one pedestrian protection airbag. In one example, pyrotechnic hood lifter can be activated first and a pedestrian protection airbag covering the windscreen, an A-pillar or similar can be activated subsequently. 16 The invention further relates to a device for activating an external safety system for protecting a pedestrian. The activating device is connected to a pre-crash sensor system and an in-crash sensor system, as defined above. The device comprises an arrangement for performing an object classification of a first signal from the pre-crash sensor system, which pre-crash sensor system comprises at least one object sensor and at least one thermal sensor, arranged to detect the temperature, or heat signature, of the object. The device also comprises an arrangement for determining a threshold for the in-crash sensor system as a function of the object classification. The device further comprises an arrangement for setting a first threshold as a function of a first signal indicating no object detected or an object classified as a pedestrian. The device also comprises an arrangement for setting a second threshold, higher than the first threshold, as a function of a first signal indicating an object classified as a non-pedestrian. Finally, the device comprises an arrangement for comparing a second signal from the in-crash sensor system with the threshold. If the comparison indicates that at least one predetermined condition is fulfilled, the device is activated if the threshold is exceeded.

According to the invention, the device can be arranged to weight the first threshold as a function of a first signal based on the confidence level of the object and thermal sensors, respectively. For instance, if the object sensor detects an object with a high confidence level, then the output signal from the thermal sensor can be used to confirm the detection of an object classified as a pedestrian. If the object sensor detects an object with a low confidence level while the output signal from the thermal sensor has a high 17 confidence level, then this indicates the detection of an object classified as a pedestrian. On the other hand, if the output signals from the object and thermal sensors have a low confidence level, then this indicates that an object classified as a non-pedestrian is detected.

Further, the device can be arranged to weight the first threshold as a function of a first signal based on a comparison between the detected object temperature and the ambient temperature. In order to differentiate a pedestrian, or a mammal, from the ambient background and from moving/inanimate objects (non-pedestrians), the thermal sensor is preferably, but not necessarily arranged to detect objects in a temperature window of approximately 30-42 °C. This temperature range covers commonly occurring core and surface temperatures for pedestrians and most mammals. In this way the output signal from the thermal sensor can be used as an additional means for filtering false positive and/or false negative indications from the one or more object sensors.

The discrimination capability of an infrared sensor system can be reduced under hot conditions or in tropical climates where background objects can have a temperature exceeding 00, e.g. in the daytime when the sun is shining. Under such conditions, the device will mainly rely on object sensors.

The device can also be arranged to fuse the output signals from the object and thermal sensors into a first signal. This fused signal forms the first signal from the pre-crash sensor 30 system used for performing an object classification.

According to the invention the at least one thermal sensor is preferably an infra-red (IR) sensor. The at least one object 18 sensor is a radar and/or a vision sensor. In this context the term "radar" is used for an arrangement detecting reflected pulses from an emitter on the vehicle, which pulses can have any suitable wavelength. The at least one vision sensor can be a suitable image detecting device, such as a camera.

According to the invention the device comprises an arrangement for determining a speed threshold relative to a nominal speed threshold for the in-crash sensor system as a function of the object classification. A device of this type can be active in a predetermined vehicle speed range, for instance between 20 km/h and 50 km/h. In this case, the nominal speed threshold is the minimum speed at which the vehicle safety system is active, that is 20 km/h.

According to a first example the device comprises an arrangement for setting the speed threshold higher than the nominal speed threshold as a function of the first signal indicating no object detected. The minimum speed at which the vehicle safety system is active is set to, for instance, km/h to make the operative range 30-50 km/h.

According to a second example the device comprises an arrangement for setting the speed threshold lower than the nominal speed threshold as a function of the first signal indicating an object classified as a pedestrian. The minimum speed at which the vehicle safety system is active is set to, for instance, 15 km/h to make the operative range 15-50 km/h.

According to a second example the device comprises an arrangement for setting the speed threshold higher than the nominal speed threshold as a function of the first signal indicating an object classified as a non-pedestrian. The 19 minimum speed at which the vehicle safety system is active is set to, for instance, 30 km/h to make the operative range 3050 km/h.

The invention also relates to a method for activating an external safety system for protecting a pedestrian. The system comprises a device for activating the external safety system, wherein the device is connected to a pre-crash sensor system and an in-crash sensor system. The device is arranged to perform an object classification of a first signal from the pre-crash sensor system. The method involves the steps of: detecting an object using at least one passive sensor, arranged to detect the temperature of the object, and at least one active sensor; - fusing the output signals from the passive and active sensors into a first signal; performing an object classification of a first signal from the pre-crash sensor system; -determining a threshold for the in-crash sensor system as a function of the object classification; setting a first threshold as a function of a first signal indicating no object detected or an object classified as a pedestrian; setting a second threshold, higher than the first threshold, as a function of a first signal indicating an object classified as a non-pedestrian; and comparing a second signal from the in-crash sensor system with the threshold, and activating the device if the threshold is exceeded.

According to the invention, the method can involve the step of weighting the first threshold (11) as a function of a first signal (Al) based on the confidence level of the passive and active sensors, respectively.

In addition, the method can perform a weighting of the first threshold as a function of a first signal based on a comparison between the detected object temperature and the ambient temperature.

According to the invention, the method can further involve determining a speed threshold relative to a nominal speed threshold for the in-crash sensor system as a function of the object classification.

According to a first example, the method involves setting the speed threshold higher than the nominal speed threshold as a function of the first signal indicating no object detected. According to a second example, the method involves setting the speed threshold lower than the nominal speed threshold as a function of the first signal indicating an object classified as a pedestrian. According to a third example, the method involves setting the speed threshold higher than the nominal speed threshold as a function of the first signal indicating an object classified as a non-pedestrian.

The method can further involve using an external safety system comprising two or more zones, provided with a pre-crash sensor system, and performing an object classification of a first signal from the pre-crash sensor system. Individual thresholds can be determined for each zone, as a function of a first signal from the pre-crash sensor system. The external safety system in each zone can be activated individually if the set threshold for the respective zone is exceeded. 10 21 Activation is performed if the comparison of the first and second signal indicates that at least one predetermined condition is fulfilled. A number of advantages are obtained by means of the present invention. The present invention provides for an algorithm to take into account the temperature of a detected object, such as a pedestrian or an animal. The output from at least one thermal sensor can be used for confirmation that an object has been detected based on sensor confidence levels and/or for filtering false positive and/or false negative indications from the one or more object sensors. In this way it is possible to precisely infer the impacting object from the impact signal, that is, the signal of an in-crash sensor system. The device according to the invention when used for activating an actuator system for protecting a pedestrian, a cyclist, or similar has the advantage that the algorithm used for activating the actuator system uses input signals from both a pre-crash sensor system and an in-crash sensor system. Advantageously, the signal of the pre-crash sensor system is used for determining a threshold level for the in-crash sensor system to generate a triggering signal for the external safety system. The signal of the in-crash sensor system is compared to a threshold determined by the pre-crash sensor system, whereupon the signal of the in-crash sensor system will determine whether to trigger the external safety system or not. According to the invention it is also possible to take into consideration the relative speed into account in evaluating the signal from the in-crash sensor system.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention will now be described more in detail with reference to the appended schematic drawings, where: 22 Figure 1 shows a schematic plan view of a vehicle provided with a safety system according to the invention; Figure 2 shows a schematic diagram indicating threshold levels for different situations; Figure 3 shows a schematic flow chart related to the processing of the sensor output signals of the safety system; Figure 4 shows a schematic plan view of a vehicle provided with an alternative embodiment of the invention. 10 DETAILED DESCRIPTION Figure 1 shows a schematic plan view of a vehicle provided with a safety system according to the invention. In Figure 1, a sensor system 8 is shown with an associated vehicle 9. The sensor system 8 is configured for a forward looking application and has the ability to sense an approaching object and prepare the vehicle 9 for an impact with a pedestrian, a cyclist or similar. In this example the application is shown with single sensors for radar, vision and temperature, respectively, but it is also possible to provide multiple sensors having overlapping fields of view.

The sensor system 8 includes a radar sensor 10 which receives a radio frequency signal, preferably in the microwave region emanating from an antenna (not shown). A radar sensor provides radar output signal 20 to an electronic control module (ECM) 12. A vision sensor 11 is preferably mounted to an upper portion of the vehicle 9, such as, along the windshield header aimed forward to provide vision information.

The vision sensor 11 provides a vision output signal 22 to the ECM 12. A thermal sensor 15 is preferably mounted to an upper portion of the vehicle 9, adjacent the vision sensor 11, to 23 provide temperature information. The thermal sensor 15 provides a thermal output signal 23 to the ECM 12. In this example, the field of view of the sensors 11, 15 is less than 180°.

The sensor system 8 further includes one or more contact sensors 16 (one shown), such as accelerometers or pressure sensors, for detecting an actual impact. The contact sensor 16 provides contact output signal 24 to the ECM 12. The contact sensors 16 are also referred to as in-crash sensors. The radar sensor 10, the vision sensor 11 and the thermal sensor 15 are also referred to as pre-crash sensors. The ECM 12 combines the radar output signal 20, the vision output signal 22, the thermal output signal 23 and the contact output signal 24 to determine if a collision is imminent and whether to adjust the threshold for the in-crash sensors.

The ECM 12 is in electrical communication with a schematically indicated external airbag 13, which can be located in the vicinity of the front bumper and/or the windshield. Control signals are also received by a hood lifter 14 from the ECM 12. The deployment timing of the external airbag 13 and the hood lifter 14 can be adjusted when an impact is detected by the in-crash sensors, thereby reducing the potential impact for a pedestrian.

Further control signals that can adjust the deployment timing of additional expandable structures (not shown) based on the radar, the vision or the thermal output and the in-crash sensor output, can be transmitted by the ECM 12, which is in electrical communication with such expandable structures. Examples of additional expandable structures include safety devices, such as, expanding bumpers or external airbags 24 covering the windshield and/or the A-pillars on either side thereof, The deployment of one or more of the expandable structures can be timed to better manage the effect of the expandable structure, in response to the type, bearing or closing velocity of an object about to impact the vehicle 9 when an impact is detected by the in-crash sensors.

Although, specific examples are provided above it is readily contemplated in accordance with the present invention, that one or all of the measurements provided byeach of the sensors may be used in adjusting various deployment characteristics of a safety device as required.

For instance, it can be desirable to trigger a number of safety devices in sequence. One example of sequentially triggered devices can be a pyrotechnical hood lifter that is triggered before a pedestrian protection airbag for the windshield, as a pedestrian will often impact the hood prior to the windshield. The order and timing of a triggering sequence is determined by the ECM 12 in response to the type, bearing and closing velocity of an object about to impact the vehicle 9 when an impact is detected by the in-crash sensors.

According to the invention, information obtained from the pre- crash sensors yield valuable data for deploying an external airbag. In order to avoid incorrect deployment of the external airbag and other parts of the safety system it is desirable to reduce "false-positive" (FP) indications and increase "true -positive" (TP) indications from the sensors and the ECM 12.

This is achieved by combining the information from the pre-crash sensors and the in-crash sensors. Information from the pre-crash sensors is used for setting a threshold for the in-crash sensor system.

The radar sensor 10 analyzes a radio frequency signal reflected off an object to obtain a range measurement, a closing velocity, and a radar cross section. A time of impact estimate is calculated based on range measurement and the closing velocity. The range measurement is the distance between the object and vehicle 9. The radar sensor 10 provides distance information with high accuracy, typically within 5 cm. The closing velocity is a measure of the relative speed between the object and the vehicle 9. The time of impact estimate is compared with the necessary time to deploy the safety device, such as an external air bag. Typically deployment time of an external airbag is between 200 ms and 30 ms. In addition, the range measurement is compared with the necessary clearance distance from the vehicle 9 to deploy the safety device. Typically clearance distance for an external air bag is between 100 mm to 800 ms. The closing velocity is also used to determine the severity of impact. High closing velocities are associated with a more severe impact, while lower closing velocities are associated with a less severe impact.

The radar cross section is a measure of the strength of the reflected radio frequency signal. The strength of the reflected signal is generally related to the size and shape of the object. The size and shape is used to access the threat of the object. The ECM 12 processes the time of impact, severity of impact, and threat assessment to provide a radar output signal.

The vision sensor 11 provides means for a vision range measurement, a bearing valve means for determining the angular deviation from a longitudinal axis through the center of the 26 vehicle, means for determining the bearing rate (the rate of change of angular deviation), and means for determining the physical size of the object. These measurements allow the ECM 12 to provide a vision sensor output signal.

The ECM 12 uses one or more of the radar, vision and thermal output signals to perform a classification of the object in order to provide a pre-crash sensor output signal indicating that; 1) a correct object has been identified, i.e. a pedestrian, cyclist, etc. an incorrect object has been identified, i.e. a non-pedestrian object; or no object has been identified.

Similarly, by processing the information from one or more contact sensors 16, such as accelerometers or pressure sensors for detecting an actual impact, the ECM 12 can provide an in-crash sensor output signal indicating that; 1) a correct object has been identified, i.e. a pedestrian, cyclist, etc. an incorrect object has been identified, i.e. a non-pedestrian object; or no object has been identified.

Figure 2 shows a schematic diagram indicating possible threshold levels for different situations. In response to the pre-crash sensor output signal, the ECM 12 described above performs a classification of a detected object in order to set a desired threshold Ti, T2 for the in-crash sensor system as a function of the object classification. 27 The pre-crash sensor output signal can be classified to indicate one of the following situations: Case 1)a correct object has been identified,i.e.a pedestrian, cyclist, etc; Case 2)an incorrect object has been identified, i.e. a non- pedestrian object; or Case 3)no object has been identified.

According to the invention, the ECM 12 is arranged to set a threshold Ti, T2 relative to a nominal threshold Tn, as a function of a first signal indicating a particular object classification. In this context, the nominal threshold Tn is a stored threshold that is used by the ECM 12 when no input is available from the active sensor systems. Further, the device is arranged to interpolate the set threshold between the nominal threshold Tn and an upper or a lower threshold limit value T1lim, 121im as a function of the probability of accurate classification for the first signal. The ECM 12 is then arranged compare a second signal A2, transmitted to the ECM 12 from the in-crash sensor system, with the set threshold Ti, T2, whereby the external safety system is activated if the set threshold Ti, T2 is exceeded.

Case 1 According to a first alternative example, the ECM 12 is arranged to set a first threshold Ti equal to a minimum threshold limit value T1lim if the probability of accurate classification has a maximum value, as a function of the first signal indicating that an object classified as a pedestrian is detected. In this case, the classification has been deemed to have a probability of 100%, or near 100% (e.g. 95-100%), and the object is classified as a pedestrian with certainty or at least with a very high degree of certainty. 28 According to an alternative first example, the ECM 12 is arranged to set a first threshold 11 equal to a minimum threshold limit value Tllim, as a function of a first signal 5 Al indicating no object detected.

According to a second alternative example, the ECM 12 is arranged to set a first threshold 11 higher than the minimum threshold limit value Tllim if the probability of accurate classification is between a minimum and a maximum value, as a function of the first signal indicating no object detected or that an object classified as a possible pedestrian is detected. In this case the ECM 12 has set a first threshold 11 interpolated between the nominal threshold In and the minimum threshold limit value Tllim, as indicated with a dashed line in Figure 2. Here the classification indicates that it is probable that an object classified as a pedestrian has been detected. Depending on the degree of probability of the classification, the first threshold 11 is interpolated as a linear function between the nominal threshold Tn and the minimum threshold limit value Tllim, as indicated by the arrow R1 in Figure 2.

Case 2 According to a third alternative example, the ECM 12 is arranged to set a second threshold 12 equal to a maximum threshold limit T2iim value if the probability of accurate classification has a maximum value, as a function of the first signal Al indicating that an object classified as a non- pedestrian is detected. In this case, the classification has been deemed to have a probability of 100%, or near 100% (e.g. 95-100%), and the object is classified as a non-pedestrian 29 with certainty or at least with a very high degree of certainty.

According to a fourth alternative example, the ECM 12 is 5 arranged to set a second threshold T2 lower than the maximum threshold limit value T2lim if the probability of accurate classification between a minimum and a maximum value, as a function of the first signal Al indicating that an object classified as a possible non-pedestrian is detected. Depending 10 on the degree of probability of the classification, the second threshold T2 is interpolated as a linear function between the nominal threshold Tn and the maximum threshold limit value T2lim, as indicated by the arrow R2 in Figure 2.

Case 3 According to a fifth alternative example the ECM 12 is arranged to set a threshold equal to the nominal threshold In, as a function of a first signal Al indicating that no input is available from the pre-crash sensor or pre-crash sensor failure. This situation can occur, for instance, for a sensor which can only detect some types of objects but does not have a generic object classification, like mono-vision system In this case, the input A2 from the in-crash sensors will be decisive for the triggering of the external safety system.

According to a further alternative example the device is arranged to set a threshold equal to a minimum threshold limit value Tllim, as a function of a first signal Al indicating no object detected. This situation can occur, for instance, for a sensor where any object impacting the car will normally be seen, such as a radar system. For such systems, a standing object in the path of the vehicle is very likely to be seen. However, because such sensors have a limited field of view a moving target passing in front of the vehicle may be hit before being detected by a sensor. In this case, it can be assumed that if the object was not seen, it is because it has moved out in front of the vehicle. Such an object is likely to be a pedestrian. Therefore a threshold equal to a minimum threshold limit value should be set.

Figure 3 provides a signal and decision flow chart related to the processing of the pre-crash sensor output signal and the in-crash sensor output signal to determine if the safety system should be activated. The process is started 30 when the ECM 12 receives multiple input signals comprising a radar output signal 20, a vision output signal 22, a thermal output signal 23 and a contact output signal 24 (see Figure 1), indicated by arrow AO. In a first step 31 the ECM 12 performs an object classification of the output signals 20, 22, 23 from the pre-crash sensor system to generate a first signal in the form of a pre-crash output signal Al. In a second step 32 the pre-crash output signal Al is used for setting a threshold for the in-crash sensor system based on whether an object is detected or not. If it is determined that a correct object or no object is detected (YES) then a first threshold Tl is set for the in-crash sensor system. The setting of the first threshold is described under the heading "Case 1" above. If it is determined that an object has been detected, but not a correct object (NO) then a second, relatively higher threshold T2 is set for the in-crash sensor system. This higher threshold T2 for the in-crash sensor system is used when the detected object is non-pedestrian. The setting of the first threshold is described under the heading "Case 2" above.

If it is determined that no input is available from the pre-crash sensors, or that a sensor failure has occurred, then the 31 ECM 12 is arranged to set a threshold equal to the nominal threshold In as described under the heading "Case 3" above.

In a third step 33 the ECM 12 processes the output signal 24 5 from the in-crash sensor system to generate a second signal in the form of an in-crash output signal A2. In a fourth step 34 the in-crash output signal A2 is combined with the pre-crash output signal Al to determine whether to activate (YES) the safety system or not (NO). The control algorithm of the ECM 12 10 evaluates the following conditions for the in-crash output signal A2 combined with the pre-crash output signal Al before activating the safety system: Example 1: If it is determined that the in-crash output signal A2 indicates that a correct object or no object is detected, when the pre-crash output signal Al has indicated that a correct object is detected (YES), then the safety system is activated 35.

Example 2: If it is determined that the in-crash output signal A2 indicates that an object has been detected, but not a correct object, when the pre-crash output signal Al has indicated that no object is detected (YES), then the safety system is activated 35.

Example 3: If it is determined that the in-crash output signal A2 indicates that a correct object is detected, when the pre-crash output signal Al has indicated that an object has been detected, but not a correct object (YES), then the safety system is activated 35.

The ECM 12 also detects that a threshold has been exceeded before activating the safety system. In the first two cases 32 the first, relatively lower threshold Ti is used for the in-crash sensor system, as the combined sensor decision is deemed to be a true positive (TP) triggering. In the third case the second, relatively higher threshold T2 is used for the in-5 crash sensor system, as the combined sensor decision is deemed to be a false positive (FP) triggering. Using a higher threshold T2 for the in-crash sensor system will increase the probability of a correct activation of the safety system. If none of the above cases are detected (NO), then the process 10 returns 36 to start 30.

The present invention provides for an algorithm to take into account the speed between the vehicle and the object, such as a pedestrian, using a pre-crash sensor system, in order to precisely to infer the impacting object from the impact signal of an in-crash sensor system. The device according to the invention when used for activating an actuator system for protecting a pedestrian, a cyclist, or similar has the advantage that the algorithm used for activating the actuator system uses input signals from both a pre-crash sensor system and an in-crash sensor system. The signal Al of the pre-crash sensor system is used for determining a threshold level T for the in-crash sensor system to generate a triggering signal for the external safety system. The signal A2 of the in-crash sensor system is compared to a threshold T1,T2 determined by the pre-crash sensor system, whereupon the signal A2 of the in-crash sensor system will be decisive when determining whether to trigger the external safety system or not. For instance, when there is no pre-crash input signal Al, e.g. due to a malfunction or if there is no object to detect, then a nominal threshold In is set. If the in-crash output signal A2 indicates that a correct object is detected (YES) then the external safety system is triggered. 33 According to the invention it is also possible to take into consideration the relative speed into account in evaluating the signal from the in-crash sensor system.

Figure 4 shows a schematic plan view of a vehicle provided with an alternative safety system according to the invention. Using the reference numerals of Figure 1, Figure 4 shows a sensor system 8 is shown with an associated vehicle 9. The sensor system 8 is configured for a forward looking application and has the ability to sense an approaching object and prepare the vehicle 9 for an impact with a pedestrian, a cyclist or similar. In this example the application is shown with single sensors for radar, vision and temperature, respectively, but it is also possible to provide multiple sensors having overlapping fields of view.

In this example, the total field of view of the sensors 11, 15 is less than 1800 and is split into two zones Z1, Z2 covering the right and left side of the vehicle, respectively. In this example, both zones Z1, Z2 extend a predetermined angle to the right and left relative to a datum line X corresponding to a central longitudinal axis through the vehicle. The sensor system 8 includes a radar sensor 10 which receives a radio frequency signal, preferably in the microwave region emanating from an antenna (not shown). The radar sensor 10 provides radar output signals 20, 20' from each zone Z1, Z2 to an electronic control module (ECM) 12. A vision sensor 11 is preferably mounted to an upper portion of the vehicle 9, such as, along the windshield header aimed forward to provide vision information. A thermal sensor 15 is preferably mounted to an upper portion of the vehicle 9, adjacent the vision sensor 11, to provide temperature information. The vision and 34 thermal sensors 11, 15 provide individual vision and temperature output signals 22, 22', 23, 23' from each zone Z1, Z2 to the ECM 12.

The sensor system 8 further includes one or more contact sensors 16, 16' for each zone Z1, Z2 (one shown), such as accelerometers or pressure sensors, for detecting an actual impact. Each zone Z1, Z2 is provided with individual contact sensors 16, 16' which provide contact output signals 24, 24' from each zone Z1, Z2 to the ECM 12. Figure 3 only shows one contact sensor per zone for reasons of clarity, but each of the indicated contact sensors 16 and 16' can represent multiple sensors distributed across the front of the vehicle. The contact sensors 16, 16' are also referred to as in-crash sensors. The radar sensor 10, the vision sensor 11 and the thermal sensor 15 are also referred to as pre-crash sensors. The ECM 12 combines radar output signals 20, 20', the vision output signals 22, 22', the thermal output signals 23, 23' and contact output signals 24, 24' to determine if a collision is imminent in one or both zones Z1, Z2 and whether to adjust the threshold for the in-crash sensors in the respective zone.

The ECM 12 is in electrical communication with a schematically indicated external airbag 13. For reasons of clarity, only one airbag 13 is shown although the safety system can comprise individually controllable airbags in the respective zones, which airbags can be located in the vicinity of the left and right portions of the front bumper, the left and right portions of the windshield. Also, inflatable structures along each A-pillar can be provided. Control signals are also received by a hood lifter 14 from the ECM 12. The deployment timing of the external airbags 13 in the respective zone Z1, Z2 and the hood lifter 14 can be adjusted when an impact is detected by the in-crash sensors, thereby reducing the potential impact for a pedestrian.

Further control signals that can adjust the deployment timing of additional expandable or inflatable structures (not shown) based on the radar, the vision or the thermal output and the in-crash sensor output, can be transmitted by the ECM 12, which is in electrical communication with such expandable structures. Examples of additional expandable structures include safety devices, such as, expanding bumpers or external airbags covering the windshield and/or the A-pillars on either side thereof, The deployment of one or more of the expandable structures can be individually timed for one or more zones to better manage the effect of the expandable structure, in response to the type, bearing or closing velocity of an object about to impact the vehicle 9 when an impact is detected by the in-crash sensors.

Although, specific examples are provided above it is readily contemplated in accordance with the present invention, that one or all of the measurements provided by each of the sensors may be used in adjusting various deployment characteristics of a safety device as required.

For instance, it can be desirable to trigger a number of safety devices in sequence. One example of sequentially triggered devices can be a pyrotechnical hood lifter that is triggered before a pedestrian protection airbag for the windshield, as a pedestrian will often impact the hood prior to the windshield. The order and timing of a triggering sequence is determined by the ECM 12 in response to the type, bearing and closing velocity of an object about to impact the vehicle 9 when an impact is detected by the in-crash sensors. 36 It can also be desirable to trigger a number of safety devices individually in different zones. The pre-crash sensor system is arranged to monitor all zones and will issue an individual 5 first signal for each zone. The external safety system in each zone can be activated individually if the set threshold for a particular zone is exceeded. The use of multiple zones allows for multiple target tracking. In this way it is possible to activate the external safety system for a detected first 10 target, such as a pedestrian, in one zone while preventing actuation for a detected second target, such as a fixed object, in a second zone.

As in the example described for Figure 1 above, information obtained from the pre-crash sensors yield data for deploying one or more external airbag. In order to avoid incorrect deployment of the external airbag and other parts of the safety system it is desirable to reduce "false-positive" (FP) indications and increase "true-positive" (TP) indications from the sensors and the ECM 12. This is achieved by combining the information from the pre-crash sensors and the in-crash sensors. Information from the pre-crash sensors is used for setting a threshold for the in-crash sensor system.

The radar sensor 10 analyzes a radio frequency signal reflected off an object to obtain a range measurement, a closing velocity, and a radar cross section. For the example shown in Figure 3, this analysis is performed for each zone Z1, Z2 and an output signal is provided for the respective zone. A time of impact estimate for the respective zone is calculated based on range measurement and the closing velocity. The range measurement is the distance between an object and vehicle 9. The radar sensor 10 provides distance 37 information with high accuracy, typically within 5 cm. The closing velocity is a measure of the relative speed between the object and the vehicle 9. The time of impact estimate is compared with the necessary time to deploy a safety device, such as an external air bag, in one or more zones. Typically deployment time of an external airbag is between 200 ms and 30 ms. In addition, the range measurement is compared with the necessary clearance distance from the vehicle 9 to deploy a safety device in one or more zones. Typically clearance distance for an external air bag is between 100 mm to 800 ms. The closing velocity is also used to determine the severity of impact. High closing velocities are associated with a more severe impact, while lower closing velocities are associated with a less severe impact.

The vision sensor 11 provides means for a vision range measurement, a bearing valve means for determining the angular deviation from a longitudinal axis through the center of the vehicle, means for determining the bearing rate (the rate of change of angular deviation), and means for determining the physical size of the object. These measurements allow the ECM 12 to provide a vision sensor output signal 22, 22' for each zone Z1, Z2. Similarly, the thermal sensor 15 provides a thermal sensor output signal 23, 23' for each zone Z1, Z2. 38 As described in connection with Figures 2A and 2B above, the ECM 12 uses one or more of the radar, vision and thermal output signals 20, 22; 20', 22'; 23, 23' to perform a classification of the object in order to provide a pre-crash sensor output signal. The object classification is performed individually for each zone Z1, Z2, based on the output signals from the pre-crash sensors and in-crash sensors covering the respective zone. For each zone, a pre-crash sensor output signal is provided, indicating that; 1) a correct object has been identified, i.e. a pedestrian, cyclist, etc. an incorrect object has been identified, i.e. a non-pedestrian object; or no object has been identified.

Similarly, by processing the information from one or more contact sensors 16 in each zone Z1, Z2, such as accelerometers or pressure sensors for detecting an actual impact, the ECM 12 can provide an in-crash sensor output signal for the respective zone indicating that; a correct object has been identified, i.e. a pedestrian, cyclist, etc. an incorrect object has been identified, i.e. a non-pedestrian object; or 25 3) no object has been identified.

Figure 2A provides a signal and decision flow chart related to the processing of the pre-crash sensor output signal and the in-crash sensor output signal to determine if the safety system should be activated, by setting suitable contact thresholds in response to the output signals. The setting of thresholds is performed individually for each zone, based on the output signals from the pre-crash sensors and in-crash 39 sensors covering the respective zone. Hence, the example shown in Figure 3 uses the same process for setting thresholds as the example shown in Figures 1 and 2A, with the difference that the process is performed simultaneously for each zone to set individual thresholds for the respective zones.

Figure 2B provides an alternative signal and decision flow chart related to the processing of the pre-crash sensor output signal and the in-crash sensor output signal. As described above, this processing involves setting of speed thresholds in addition to the contact thresholds. The setting of contact and speed thresholds is performed individually for each zone, based on the output signals from the pre-crash sensors and in-crash sensors covering the respective zone. Hence, the example shown in Figure 4 uses the same process for setting thresholds as the example shown in Figures 1 and 2B, with the difference that the process is performed simultaneously for each zone to set individual thresholds.

The invention is not limited to the embodiments described above, but can be varied within the scope of the claims. For instance, the examples given for the operative speed range and the suggested minimum speed threshold

Claims

1. A vehicle safety system comprising an external safety system for protecting a pedestrian, and a device for activating the external safety system the device being connected to a pre-crash sensor system (10, 11, 15) and an in-crash sensor system (16), the device being arranged to perform an object classification of a first signal from the pre-crash sensor system;characterized in that the device is arranged to determine a threshold (Ti, T2) for the in-crash sensor system as a function of the object classification; that the pre-crash sensor system comprises at least one object sensor (10, 11), and at least one thermal sensor (15), 15 arranged to detect the temperature of the object; where the device is arranged to set a first threshold (Ti) as a function of a first signal (Al) indicating no object detected or an object classified as a pedestrian; and to set a second threshold (12), higher than the first 20 threshold (T1), as a function of a first signal (Al) indicating an object classified as a non-pedestrian; and that the device is arranged to compare a second signal from the in-crash sensor system with the threshold, whereby the external safety system is activated if the set threshold (Ti, T2) is exceeded.

2. A vehicle safety system according to claim 1, characterized in that the device is arranged to weight the first threshold (Ti) as a function of a first 30 signal (Al) based on the confidence level of the object and thermal sensors (10, 11, 15), respectively. 41

3. A vehicle safety system according to claim 1 or 2, characterized in that the device is arranged to weight the first threshold (Ti) as a function of a first signal (A1) based on a comparison between the detected object 5 temperature and the ambient temperature.

4. A device according to any one of claims 1-3, characterized in that the device is arranged to the device is arranged to fuse the output signals from the object and thermal sensors (10, 11, 15) into a first signal (A1).

5. A vehicle safety system according to any one of claims 1-4, characterized in that the thermal sensor (15) 15 is an infra-red sensor.

6. A vehicle safety system according to any one of claims 1-5, characterized in that the at least one object sensor is a radar and/or a vision sensor (10, 11).

7. A device for activating an external safety system for protecting a pedestrian, the device being connected to a pre-crash sensor system (10, 11, 15) and an in-crash sensor system (16), the device, comprising: - an arrangement for performing an object classification of a first signal (A1) from the pre-crash sensor system (10, 11, 15), which pre-crash sensor system comprises at least one object sensor (10, 11) and at least one thermal sensor (15), arranged to detect the temperature of the object; - an arrangement for determining a threshold (Ti, T2) for the in-crash sensor system (16) as a function of the object classification; 42 1. an arrangement for setting a first threshold (11) as a function of a first signal (Al) indicating no object detected or an object classified as a pedestrian; 2. an arrangement for setting a second threshold (12), higher than the first threshold (T1), as a function of a first signal (Al) indicating an object classified as a non-pedestrian; and 3. an arrangement for comparing a second signal (A2) from the in-crash sensor system (16) with the threshold (11, 12), the device being activated if the threshold is exceeded.

8. A device according to claim 7, characterized in that the device is arranged to weight the first threshold (11) as a function of the first signal (Al) based on the confidence level of the object and thermal sensors, respectively.

9. A device according to claim 7 or 8, characterized in that the device is arranged to weight the first threshold (11) as a function of the first signal (Al) based on a comparison between the detected object temperature and the ambient temperature.

10. A device according to any one of claim 7-9, characterized in that the device is arranged to fuse the output signals from the object and thermal sensors (10, 11, 15) into a first signal (Al)

11. Method for activating an external safety system for protecting a pedestrian which system comprises a device for activating the external safety system, the device being connected to a pre-crash sensor system (10, 11, 15) and an in-crash sensor system (16) and being arranged to perform an 43 object classification of a first signal from the pre-crash sensor system; the method involving the steps of: 1. detecting an object using at least one object sensor (10, 11), and at least one thermal sensor (15), arranged to detect the temperature of the object; 2. fusing the output signals from the object and thermal sensors (10, 11, 15) into a first signal (Al); 3. performing an object classification of a first signal (Al) from the pre-crash sensor system (10, 11, 15); - determining a threshold (Ti, 12) for the in-crash sensor system (16) as a function of the object classification; 4. setting a first threshold (Ti) as a function of a first signal (Al) indicating no object detected or an object classified as a pedestrian; - setting a second threshold (12), higher than the first threshold (Ti), as a function of a first signal (Al) indicating an object classified as a non-pedestrian; and 5. comparing a second signal (A2) from the in-crash sensor system (16) with the threshold (Ti, 12), and activating the device if the threshold is exceeded.

12. Method according to claim 11, characterized by weighting the first threshold (11) asafunction of the first signal (Al) based on the confidence level of the 25 object and thermal sensors (10, 11, 15), respectively.

13. Methodaccordingtoclaim11or12, characterized by weighting the first threshold (Ti) as a function of the first signal (Al) based on a comparison between the detected object temperature and the ambient temperature. Ansokningsnummer: 1450739-6 I denna ansokan saknades patentkrav vid ingivningen. 1:6 Z.6!d —7- wozi. 3/4