Method for controlling a vehicle when an obstacle is detected in surroundings of the vehicle; control device for a vehicle with an autonomous driving function; computer readable medium and motor vehicle

Abstract

The invention is concerned with a method for controlling a vehicle (10) when an obstacle (18) is detected in surroundings (39) of the vehicle (10), wherein an autonomous driving function of the vehicle (10) plans a trajectory (15). An observer function (23) determines and/or adapts a size of a dynamic protection zone (26) that extends in a current driving direction (13), wherein the size depends on a current driving speed of the vehicle (10), wherein the observer function determines whether the obstacle (18) is detected within the dynamic protection zone (26), and in the case the obstacle (18) is detected, a limiter function (29) is signaled by the observer function, wherein the limiter function (29) is provided in the signal path (30) and the limiter function (29) reduces speed values of the planned trajectory (15) without influencing the line (25) of the movement of the vehicle (10), thereby causing the observer function (23) to shrink the dynamic protection zone (26) until the obstacle (18) lies outside the shrunken dynamic protection zone (26).

Claims

1. Method for controlling a vehicle when an obstacle is detected in surroundings of the vehicle, wherein an autonomous driving function of the vehicle plans a trajectory comprising a line of movement and corresponding speed values along the line of movement and sends the planned trajectory to a driving actuator over a signal path, characterized in that an observer function of a control device of the vehicle determines and/or adapts a shape and/or size of a dynamic protection zone that extends in a current driving direction from the vehicle, wherein the shape and/or size depends on a current driving speed of the vehicle, in that the dynamic protection zone is dynamically expanded if the driving speed of the vehicle is increased and the dynamic protection zone is shrunk if the driving speed of the vehicle is reduced, wherein the observer function determines whether the obstacle is detected within the dynamic protection zone, and in the case the obstacle is detected within the dynamic protection zone, a limiter function is signaled by the observer function, wherein the limiter function is provided in the signal path and, if signaled by the observer function, the limiter function reduces the speed values of the planned trajectory without influencing the line of the movement of the vehicle resulting in a limited trajectory, thereby causing the observer function to shrink the dynamic protection zone until the obstacle lies outside the shrunken dynamic protection zone.

2. The method according to claim 1, wherein the observer function detects the obstacle based on sensor data that indicate the presence of an object and the object is classified as an obstacle, if the physical size of the object is at least a predefined minimum size.

3. The method according to claim 2, the observer function detects the obstacle independently of a type of the object.

4. The method according to claim 1, wherein the observer function classifies an object as an obstacle if a classifier function of the control device and/or the autonomous driving function signals that a certainty value regarding a classification result for the object is within a predefined interval.

5. The method according to claim 1, wherein the observer function detects the obstacle based on at least a sensor set comprising one or more sensors, and wherein a) the sensor set of the observer function comprises at least one sensor that is also comprised in a sensor set of the autonomous driving function and/or b) the sensor set of the observer function comprises at least one sensor that is different from each sensor in the sensor set of the autonomous driving function.

6. The method according to claim 1, wherein the protection zone is wider than a width of the vehicle such that the protection zone also encompasses an obstacle that the vehicle would pass by on the basis of the planned trajectory.

7. The method according to claim 6, wherein the width of the protection zone is set on the basis of a predefined statistic distribution function of pedestrian speeds and/or at least one predefined walking speed value that describes a walking speed of a potential pedestrian who is moving perpendicular to the line of movement of the trajectory.

8. The method according to claim 7, wherein at least two predefined walking speed values are set, wherein each walking speed value is valid for different distance intervals with regard to the vehicle.

9. The method according to claim 1, wherein changing the trajectory describes a reduction of the speed of the vehicle to a speed value larger than 0 such the vehicle continues driving along the line of movement.

10. The method according to claim 1, wherein the speed values are reduced until the protection zone is shrunk to a size such that the obstacle is positioned at a border of the protection zone.

11. The method according to claim 1, wherein the vehicle is stopped by the limiter function, if a protection zone with a size of zero is required due to the position of the obstacle.

12. The method according to claim 1, wherein the shape of the protection zone is adapted to fulfill acceptance criteria of pedestrian injury rates of a given severity level, and in particular wherein the assumed injury rates influencing the shape of the protection zone consider the local and temporal expected rate of crossing pedestrians.

13. A control device for a vehicle with an autonomous driving function, the control device comprising a processing unit, wherein the processing unit is adapted to perform a method according to claim 1.

14. A computer readable data storage medium containing a programming code that comprises programming instructions for performing a method according to claim 1 when executed by a processing unit.

15. A motor vehicle comprising a sensor set and an autonomous driving function coupled to the sensor set and a driving actuator coupled to the autonomous driving function over a signal path, characterized in that a control device according to claim 13 provides a limiter function in the signal path and the limiter function is coupled to an observer function.

Description

[0034] In the following an exemplary implementation of the invention is described. The figures show:

[0035] FIG. 1 a schematic illustration of an embodiment of the inventive motor vehicle;

[0036] FIG. 2 a schematic illustration of a protection zone; and

[0037] FIG. 3 a sketch showing the vehicle of FIG. 1 while it is adapting the protection zone.

[0038] The embodiment explained in the following is a preferred embodiment of the invention. However, in the embodiment, the described components of the embodiment each represent individual features of the invention which are to be considered independently of each other and which each develop the invention also independently of each other and thereby are also to be regarded as a component of the invention in individual manner or in another than the shown combination. Furthermore, the described embodiment can also be supplemented by further features of the invention already described.

[0039] In the figures identical reference signs indicate elements that provide the same function.

[0040] FIG. 1 shows a vehicle 10 that can be a passenger vehicle or a truck. The vehicle 10 may comprise an autonomous driving function 11 for controlling a driving actuator 12 such that vehicle 10 may drive along a driving direction 13 without the help of a driver. The autonomous driving function 11 may control the driving actuator 12 by means of trajectory data 14 for setting or defining a trajectory 15 along which the driving actuator 12 shall steer the vehicle 10 and set the speed value of the driving speed of the vehicle 10.

[0041] For calculating the trajectory data 14, the autonomous driving function 11 may receive sensor data 16 from a sensor set 17. The sensor set 17 may comprise, for example, at least one camera and/or at least one radar and/or at least one lidar sensor. The sensor data 16 may describe the environment or surroundings of the vehicle 10. The autonomous driving function 11 may be implemented as a software or programming instructions for at least one microprocessor of a processing unit. The autonomous driving function 11 may receive the sensor data 16 and calculate an appropriate trajectory 15 that may lead the vehicle 10 around any obstacle described by the sensor data 16.

[0042] For this, a certain amount of calculation time is needed. Thus, if a new obstacle appears in the driving direction 13 of the vehicle 10, there may not be enough time to re-calculate the trajectory 15 for considering the new obstacle 18. If the obstacle 18 is a pedestrian 19 who walks towards a strip 20 that defines the space needed by the vehicle 10 for moving along the trajectory 15, and if the walking speed 21 is large enough to reach the strip 20 before the vehicle 10 has passed by, this could lead to a collision of the vehicle 10 with the obstacle 18.

[0043] In vehicle 10 a checker functionality 22 is provided to support the autonomous driving function 11 in reacting to a newly appearing obstacle 18 that has not been considered in the trajectory data 14 or that has been considered in the trajectory data 14 assuming a different behavior than is actually observed.

[0044] An observer function 23 may receive sensor data 24 from a sensor set 17 that may comprise the same and/or different sensors as compared to sensor set 17 for the autonomous driving function 11. On the basis of the sensor data 24 and the trajectory data 14 the observer function 23 may determine which line 25 of movement will result from the current trajectory data 14 and at which speed the vehicle 10 should move along that line 25.

[0045] The observer function 23 may define a shape of a projection zone 26 that may consider obstacles 18, like pedestrians 19, who move with a given or defined maximum speed value 27 of their walking speed 21 towards the line 25 such that they may reach the strip 20 before the vehicle 10 arrives or has passed by the obstacle 18. If an obstacle 18 is detected inside the protection zone 26, the observer function 23 may generate a limitation signal 28 for a limiter function 29. The limiter function 29 may be provided within a signal path 30 that transmits the trajectory data 14 from the autonomous driving function 11 towards the driving actuator 12. If the limiter function 29 receives the limitation signal 28, the limiter function 29 manipulates the trajectory data 14 in order to generate adapted trajectory data 31 that describe an adapted trajectory 32. The adapted trajectory data 31 set the adapted trajectory 32 such that it describes the same line of movement 25 but the speed values are limited to a maximum allowable value and/or they are reduced as compared to the speed values contained in the original trajectory data.

[0046] The observer function 23 defines a size and/or shape of the projection zone 26 as a function of the actual moving speed or driving speed of the vehicle 10. The protection zone 26 may thus be adapted in size and/or shape as a function of the driving speed of the vehicle 10 along the line of movement 25. The larger the speed, the larger the protection zone 26, which considers the longer distance needed for breaking or halting the vehicle 10 and the longer time to come to a stop and thus the more time for pedestrians to reach the line 25, thus a larger lateral extension is advantageous.

[0047] When the driving speed is reduced (as the driving actuator 12 reacts to the adapted trajectory data 31), this may change the observation that the obstacle 18 is within the protection zone 26. The protection zone 26 may have shrunk to a size small enough to leave the obstacle 18 outside or at least at a border of the protection zone. Then, no further reduction of the speed values of the trajectory data 14 is necessary and the vehicle 10 may drive past the obstacle 18 safely at that driving speed.

[0048] FIG. 2 illustrates a possible definition of the protection zone 26. The protection zone 26 may extend from the vehicle 10 along the driving direction 13. What is shown is an axis for the distance value d starting from the vehicle 10 and measured along the line 25 of movement. If the line 25 of movement is bent for driving a curve (not shown), the shape of the protection zone 26 can be adapted accordingly, i.e. it may be bent.

[0049] The protection zone 26 may comprise of three different distance intervals 33, 34, 35 of different widths, but this number of intervals is only exemplary and it may also be less or more than three. The width is measured perpendicular to the line of movement 25. The distance intervals 33 closest to the vehicle 10 may provide the broadest part of the protection zone 26 in that it considers obstacles 18 that have the largest or fastest walking speed 21 towards the line 25 of movement of the trajectory 15. This results in the broadest width 37. The next distance interval 34 may consider obstacles with a smaller walking speed resulting in a smaller width 37. The 3rd distance interval 35 may only consider obstacles that are already within the strip 20. However, this is only an example shape. There does not have to be a broadest part closest to the vehicle. As an exemplary alternative, a rectangle shape could also be provided, but the choice is not limited to the shapes shown.

[0050] FIG. 3 illustrates measurement data or sensor data 24 as a map 38 of surroundings 39 of the vehicle 10. As can be seen, the sensor data 24 describes several objects 40 that, however, are not classified as obstacles as they are away far enough, i.e. outside the protection zone 26. Behind the objects 40 a respective shadow area or shadow zone 41 results from the fact that the sensor set 17 with its sensors cannot detect any further objects behind the detected objects 40.

[0051] If an object is within the protection zone 26, such an object can become an obstacle 18, if from the sensor data 16, 24 it is clear that the object shall not be hit by the vehicle 10, as it is, e.g., a pedestrian or at least it is recognized as an object larger than a predefined minimum size. An object may also be classified as an obstacle 18, if it cannot be guaranteed or certified that this object is not or cannot hide, for example, a pedestrian. If a predefined uncertainty remains from the sensor data 24, the presence of an obstacle 18 is preferably assumed. This can be the case, if the size of the object is larger than a predefined minimum size and/or if an object recognizer, for example an artificial neural network, indicates that a classification of the object can only be provided with a certainty value that is smaller than a predefined minimum certainty value.

[0052] If an obstacle 18 is detected inside the protection zone 26, the observer function 23 will generate the limitation signal 28. Limitation signal 28 can be a binary signal only indicating that the current trajectory data 14 are not okay NOK. The limitation signal 28 can additionally or alternatively comprise a new speed value that shall be set or enforced by the limitation function 29. The new speed value can indicate at which speed the protection zone 26 will be small enough such that the obstacle 18 will not be inside the then resulting shrunken protection zone 42.

[0053] The limitation function 29 may take back the limitation, i.e. it may increase the upper limit such that trajectory data 14 with higher speed values are again passed through un-changed, when the vehicle 10 has passed by the obstacle 18.

[0054] Generating the limitation signal 28 for reducing the speed values in the trajectory data 14 by the limiter function 29 and ending the limitation signal 28 may be a dynamic process such that the projections zone 26 may be continuously pumping, i.e. alternating its shape and/or size (shrinking, growing, shrinking, growing et cetera), as the limiter function 29 and the observer function 23 interact. Whenever an obstacle 18 is inside the protection zone 26, observer function 23 may generate the limitation signal 28 such that the limitation function 29 is activated and reduces the speed values in the trajectory data 14. The upper limit that may be applied by the limiter function 29 may be gradually lowered such that the vehicle 10 decelerates gradually. This deceleration will result in a shrinking protection zone 42. Once no obstacle 18 is detected inside the shrunken protection zone 42 anymore, the observer function 23 will stop sending the limitation signal 28 such that the limiter function 29 will increase the upper limit again. This can be implemented as a gradual increase. The vehicle 10 will be accelerated accordingly until either a new obstacle 18 of the same obstacle 18 is detected inside the resulting protection zone 26 again or until the original speed value from the un-changed trajectory data 14 is reached again.

[0055] The idea described is to detect a pedestrian in a speed-dependant dynamic protection zone or a virtual area in front of the vehicle on the basis of data analysis of the sensor data (e.g. checking if the detected object can be a pedestrian based on its dimensions). The size of the protection zone is adapted to expand with regard to an increase in the speed of the motor vehicle. If the detected object cannot be guaranteed not to be a pedestrian based on the data analysis, the safety mechanism may reduce the vehicle speed or even stop the vehicle in extreme conditions. In such a case, the autonomous driving function of the vehicle may then plan another trajectory of the vehicle, along which the safety mechanism may allow a speed greater than zero. But the described logic does not depend on such a re-planning, as it simply limits the maximum speed that an autopilot may execute.

[0056] The checker functionality 22 for obstacles like crossing pedestrians (CCPchecker for crossing pedestrians) is a safety mechanism to prevent insufficient deceleration of vehicle in automated driving mode in front of crossing pedestrians. Statistic information about the speed distribution of crossing pedestrians and accident statistics of the probability of accident severities at various impact speeds between vehicles and pedestrians can be used to parametrize a dynamic protection zone in front of the vehicle with minimal dimensions.

[0057] The CCP may give protection for pedestrians that walk along the street or cross with reasonable speed resulting in low probabilities for any injury. Pedestrians cross streets with higher speed only with low probability. If they still do so, the CCP will protect against collisions with higher impact speed and consequently more severe injuries. Pedestrians that move with normal walking speed should have a very low probability of injuries implemented by low collision speed. In total, the acceptance criteria in terms of limits for the rates of pedestrian injuries at different levels of severity have to be kept.

[0058] The CCP requires only minimal classification certainty. Any detected object within the protection zone that could be a pedestrian or that could hide a pedestrian leads to a deceleration of the vehicle or even an emergency brake.

[0059] An Automated Driving System without a checker generally complies a sensor set, processing for calculating a trajectory, actuation to follow the trajectory. A checker adds an own sensor set, a safety observer and a limiter into this architecture. (Variants without own sensors using preprocessed data of the perception part of the mission channel are also possible). The observer of the checker contains a metric to decide, based on the sensor input, if a trajectory is safe or not. As long as no safety violation is detected, the limiter lets the trajectory pass through unchanged. If the observer notifies a safety violation to the limiter the limiter modifies the trajectory to stay within a safety envelope.

[0060] The CCP can use an emergency brake limiter that always forwards the path of a given trajectory but reduces the speed with maximum brake force considering the dynamic limits and the limitation of the surface friction. The CCP could also pick between several offered trajectories with different levels of violation of the protection zone.

[0061] A functionally redundant set of sensors can be used with maximum overlap between the sensors in the short range region with low opening angle directly in front of the vehicle to reach a maximum safety integrity level with respect to detection of objects that can be or can hide a pedestrian.

[0062] The checker for crossing pedestrians (CCP) allows a quantitative statement about the safety of the system with respect to crossing pedestrians. Upper limits for the rates of injuries with given severity can be calculated. Or alternatively, the lateral extension of the safety zone can be minimized for given acceptance criteria regrading acceptable injury rates and severities. The protection zone or safety zone is dynamically calculated at runtime as a function of the planned trajectory. The CCP has low complexity as it does not require pedestrian classification, advanced prediction, or own path planning and thus can be easily implemented with high safety integrity. An ASIL (Automotive Safety Integrity Level) decomposition between the mission channel that generates the original trajectory and the CCP allows to implement the mission channel with lower safety integrity.

[0063] The longitudinal extension of the protection zones may be given in terms of collision speed (after maximum brake action) and the absolute dimension may be speed dependent. Accordingly, the lateral extension of the protection zone may be given by the ratio of assumed pedestrian speed to vehicle speed.

[0064] Overall, the example shows how a checker for crossing pedestrians may be superimposed on an autonomous driving function.