阅读说明：本技术 用于识别传感器的减退的性能的方法、控制器和存储介质 (Method, controller and storage medium for identifying degraded performance of a sensor ) 是由 M·R·埃韦特 于 2019-07-31 设计创作，主要内容包括：本发明涉及一种用于识别尤其车辆中的至少一个传感器的减退的性能的方法,其中,计算所述车辆的额定轨迹；由所述车辆行驶或通过人工智能模拟所述额定轨迹；将由所述车辆行驶的实际轨迹或模拟出的轨迹与所述额定轨迹比较；在确定所述实际轨迹或所述模拟出的轨迹与所述额定轨迹存在偏差的情况下,通过控制器检查所述至少一个传感器的性能,在该性能的情况下交替地停用所述车辆的每个传感器并且借助至少一个选择的传感器由所述车辆行驶所述额定轨迹或者通过人工智能模拟所述额定轨迹。此外,本发明还涉及一种控制器、计算机程序以及机器可读的存储介质。(The invention relates to a method for detecting a reduced performance of at least one sensor, in particular in a vehicle, wherein a target trajectory of the vehicle is calculated; simulating the nominal trajectory by the vehicle or by artificial intelligence; comparing an actual trajectory or a simulated trajectory traveled by the vehicle with the nominal trajectory; if a deviation of the actual or simulated trajectory from the target trajectory is determined, the performance of the at least one sensor is checked by a controller, in which case each sensor of the vehicle is deactivated alternately and the target trajectory is driven by the vehicle by means of at least one selected sensor or simulated by artificial intelligence. The invention further relates to a controller, a computer program and a machine-readable storage medium.)

1. A method (20) for detecting a reduced performance of at least one sensor (6,8), in particular in a vehicle (2),

-calculating (22) a nominal trajectory of the vehicle (2),

-driving (23) the nominal trajectory by the vehicle (2) or simulating (24) the nominal trajectory by artificial intelligence,

-comparing (25) an actual trajectory or a simulated trajectory traveled by the vehicle (2) with the nominal trajectory,

-checking (27), by means of a controller (10), the performance of the at least one sensor (6,8) in the event of a deviation of the actual trajectory or the simulated trajectory from the nominal trajectory being determined (26), in which case each sensor (6,8) of the vehicle (2) is alternately deactivated and the nominal trajectory is driven by the vehicle (2) by means of at least one selected sensor (6,8) or is simulated by means of the artificial intelligence.

2. The method according to claim 1, wherein at least one sensor (6,8) having degraded performance is deactivated by the controller (10).

3. The method according to claim 1 or 2, wherein the method for identifying reduced performance of the at least one sensor (6,8) is performed by artificial intelligence alone or by existing artificial intelligence for an automatable driving pattern.

4. A method according to any one of claims 1 to 3, wherein the vehicle (2) is placed in a safe state in case it is determined that the actual trajectory deviates from the nominal trajectory.

5. Method according to one of claims 1 to 4, wherein the nominal trajectory is calculated inside the vehicle by means of at least one controller (10) or outside the vehicle by means of a server unit (4), wherein the nominal trajectory outside the vehicle is transmitted to the vehicle (2) via a vehicle-to-X communication connection (16).

6. The method according to any one of claims 1 to 5, wherein the actual trajectory is plausibly verified by at least one artificial intelligence inside or outside the vehicle.

7. The method according to any one of claims 1 to 6, wherein sensor data within a defined period of time is intermediately stored in a memory unit (12) and subsequently analyzed by the artificial intelligence for the purpose of performing plausibility verification.

8. The method of any of claims 1-7, wherein sensor data is analyzed by the artificial intelligence in parallel with the automated driving mode implemented.

9. A controller (10) for coupling to at least one sensor (6,8) and for evaluating sensor data, wherein the controller (10) is set up for carrying out all the steps of the method (20) according to one of the preceding claims.

10. A server unit (4) having at least one memory (18), wherein the server unit (4) is configured for connecting with at least one vehicle (2) and to perform all the steps of the method (20) according to any one of claims 1 to 8.

11. A computer and program comprising instructions which, when said computer program is executed by a computer or controller (10), arrange the computer to carry out the method according to any one of claims 1 to 8.

12. A machine-readable storage medium (12,18) on which a computer program according to claim 11 is stored.

Technical Field

The invention relates to a method for detecting a reduced performance of at least one sensor, in particular in a vehicle

And to a controller configured for carrying out the method. The invention further relates to a computer program and a machine-readable storage medium.

Background

The present invention relates to a method for automatically operating a vehicle, in particular a motor vehicle. The automatically operable vehicle may be an autonomous or partially autonomous vehicle, for example. The surroundings of the vehicle can be detected and monitored by means of sensors, as a result of which automated navigation and control of the vehicle is possible.

Over the course of a vehicle's life, sensors may deteriorate in their performance and accuracy of recognition of the vehicle's environment. For example, protective lenses in front of a camera may become blurred or discolored with aging. Furthermore, the driven mirror, for example a radar sensor, can no longer rotate with the set accuracy over time. Today, in the event of an internal sensor failure, the sensor sends a failure report to the superordinate controller. In this way, the sensor can be switched off and/or replaced in a timely manner.

If the sensor is still working, but has degraded performance or accuracy, then the monitor inside the sensor will not generate a fault report. This is caused by the fact that the sensors can also operate in partial function, whereby these sensors are not recognized as noticeable on the inside. However, the performance is becoming worse and worse, which may be endangered for performing automated driving functions.

Disclosure of Invention

The object on which the invention is based may be to provide, in particular, a method by means of which a decrease in the performance of a sensor can be detected in a timely manner.

This object is achieved by the method, the controller, the computer program and the machine-readable storage medium according to the invention. Advantageous configurations of the invention are the respective preferred embodiments.

According to one aspect of the invention, a method for identifying a reduced performance of at least one sensor, in particular in a vehicle, is provided. In one step, a nominal trajectory of the vehicle is calculated. The target trajectory is simulated by vehicle travel or by artificial intelligence. Next, the actual trajectory or the simulated trajectory traveled by the vehicle is compared with the target trajectory. If a deviation of the actual or simulated trajectory from the target trajectory is determined, the controller checks the performance of the at least one sensor, in which case each sensor of the vehicle is deactivated alternately and the target trajectory is driven by the vehicle by means of the at least one selected sensor or is simulated by means of artificial intelligence.

In one step, a nominal trajectory of the vehicle is simulated by artificial intelligence. The target trajectory is then driven by the vehicle. The actual trajectory traveled by the vehicle is compared with the simulated target trajectory. In particular, the sensor data of the at least one sensor can be read by the controller and used to perform the comparison. If a deviation of the actual trajectory from the target trajectory is determined, the controller checks the performance of the at least one sensor, in which case each sensor of the vehicle is alternately deactivated and the target trajectory is traveled over by the vehicle by means of the at least one selected sensor.

According to a further aspect of the invention, a controller for coupling to at least one sensor and for evaluating sensor data is provided, wherein the controller is set up for carrying out all the steps of the method according to the invention.

Furthermore, according to an aspect of the invention, there is provided a computer program comprising instructions which, when executed by a computer or controller, arrange the computer to perform the method of the invention.

According to a further aspect of the invention, a server unit having at least one memory is provided, wherein the server unit is configured for connecting to at least one vehicle and for carrying out all the steps of the method according to the invention.

According to another aspect of the invention, a machine-readable storage medium is provided, on which the computer program of the invention is stored.

The vehicle may be an automatically operable vehicle. Such vehicles, in particular partially automated, highly automated or fully automated vehicles, have a plurality of sensors. Such vehicle sensors may be camera sensors, radar sensors, lidar sensors, ultrasonic sensors, and the like. In particular, the sensor may be an environmental sensor which can be used to identify the environment of the vehicle and to enable vehicle control within the environment.

Furthermore, airbag sensors and ESP sensors, such as rotational speed sensors, acceleration sensors or pressure sensors, can also be read and monitored by the controller.

In vehicles, it may be necessary, in particular in terms of safety-critical functions, to identify an environmental sensor with degraded performance before it generates an internal fault or fails completely, in order to avoid a serious accident. This identification of environmental sensors in a vehicle that have degraded in their performance can be performed by the method of the invention.

The at least one sensor transmits its sensor data within the vehicle to a superordinate control unit, which can be responsible for carrying out at least one autonomous driving function within the vehicle. For example, the controller may have one or more artificial intelligence for performing automated driving functions or autonomous driving. The environmental sensors may be monitored continuously or as needed for their performance.

The sensor preferably has a basic function, which is, however, reduced, or limited compared to the original state. This may be, for example, a smaller effective range, a slower detection rate, a lower resolution or signal offset, etc. In this case, the monitor inside the sensor is not noticeable, since the sensor can still be operated with partial functionality, whereby the sensor appears inconspicuous inside. However, this deteriorated performance is endangered for performing the autonomous driving function.

According to one embodiment, the method for detecting reduced performance of the environmental sensor is performed using artificial intelligence that has been used for automated driving. The performance of environmental sensors is monitored by means of artificial intelligence that has been used for automated driving. The sensor data are processed by corresponding artificial intelligence. For example, a planned trajectory or a control of the vehicle actuators as an output of artificial intelligence can be output.

The artificial intelligence calculates a target trajectory which is to be traveled over by the vehicle in a subsequent step as accurately as possible. Furthermore, actuator manipulation can be simulated in parallel within artificial intelligence. Based on the simulation or target trajectory, the value of the output of the artificial intelligence is calculated, for example as a control of the actuator. These calculated values at the artificial intelligence output are then compared with the setpoint values or the sensor data of the sensor. For example, it may be found on another artificial intelligence inside the autonomous vehicle when simulating the trajectory, where the vehicle is expected to be after execution of the planned trajectory. If the planned trajectory is completely untrustworthy, since the vehicle is no longer located on the road after the execution of the corresponding trajectory or collides with another traffic participant or infrastructure, it is assumed that the at least one environmental sensor of the autonomous vehicle no longer functions without any defects.

Next, it is possible to identify: which of these sensors in the vehicle has degraded performance. For this purpose, weather data can be retrieved from the cloud and transmitted to the vehicle, for example, via a vehicle-to-X communication connection. In case of poor weather conditions, the performance of the camera may be insufficient, for example. The camera may not be used for autonomous driving in this scenario in principle. This fault may be present, for example, in a further environmental sensor of the vehicle, for example a radar sensor or a lidar sensor. In a further step, all the surroundings sensors are deactivated by the control unit or the vehicle, wherein the vehicle can continue to be operated at a low speed, for example at the edge of the road. Furthermore, it is also possible to study the calculated target trajectory in a simulated manner before execution on the actuator and to determine the expected change in position of the vehicle by means of artificial intelligence. Thus, sensors that degrade in their performance can be identified similar to a depth search (Tiefensuche), in which case each sensor describes a path. If the change in position of the vehicle is plausible, the deactivated environmental sensors are reactivated and the process is repeated. If a large positional deviation between the simulated or target trajectory and the actual trajectory occurs, which cannot be verified, a fault is present in the reactivated environmental sensor.

According to one embodiment, at least one sensor having reduced performance is deactivated by the controller. Corresponding sensors that do not provide the required performance may be permanently deactivated after the recognition is completed. Multiple environmental sensors may also be defective at the same time, however the probability of this being the case is low. Nevertheless, other environmental sensors may also be tested. Subsequently, the autonomous driving function of the vehicle can be continued, if possible, by means of the remaining sensors. This can be performed, for example, until a factory is reached, in which a defective environmental sensor can be replaced. The vehicle and/or controller may have different artificial intelligence that is in charge of autonomous driving or designed with different numbers or types of environmental sensors. In the event of a radar sensor failure, for example, artificial intelligence saved in the vehicle can be implemented, which is based solely on the camera and/or the lidar sensor.

According to another embodiment, at least one sensor is monitored for performance by artificial intelligence alone or by existing artificial intelligence for an automatable driving mode. The performance of environmental sensors can be monitored by means of artificial intelligence that has been used for automated driving. The sensor data are processed by corresponding artificial intelligence. Alternatively or additionally, another artificial intelligence, for example implemented into the controller, may be saved on the vehicle for performing the method. For monitoring purposes, additional artificial intelligence may be run in parallel with the artificial intelligence for autonomous driving.

According to a further embodiment, the vehicle is placed in a safe state if a deviation of the actual trajectory from the setpoint trajectory is determined. Since the artificial intelligence for autonomous driving has basically been tested and learned on the basis of a very large amount of input data, internal faults within the artificial intelligence can be excluded first. Alternatively or additionally, the vehicle may be transferred into a safe state instead of executing the trajectory after the simulation is executed. This can be performed, for example, by redundant sensors, such as GNSS sensors.

According to a further embodiment, the target trajectory is calculated inside the vehicle by means of at least one controller or outside the vehicle by means of a server unit, wherein the target trajectory outside the vehicle is transmitted to the vehicle via a vehicle-to-X communication link. Vehicle-to-vehicle communication is understood to be the exchange of information and data between vehicles. The purpose of this data exchange is to report critical and dangerous situations to the driver early. The vehicle concerned collects data such as ABS intervention, steering angle, position, direction and speed and transmits these data to the other traffic participants by radio (according to the WLAN standard, UMTS standard, LTE standard, etc.). In this case, the "line of sight" of the driver is to be extended by means of the electronic device. Vehicle-to-infrastructure is understood to be the exchange of data between a vehicle and the surrounding infrastructure (e.g. optical signal devices). Said technique is based on the co-action of sensors of different traffic partners. The two possibilities mentioned can equally be considered for the vehicle-to-X communication connection.

Additionally or alternatively, corresponding artificial intelligence for degrading (degradieren) vehicle characteristics may be downloaded from the cloud or server unit and performed in place of traditional artificial intelligence on the vehicle. Such artificial intelligence may be saved, for example, within the cloud for vehicle modification, and is preferably obtained by the vehicle through a vehicle-to-X communication connection after the vehicle has reported the defective sensor to the cloud.

In another embodiment, the actual trajectory is validated for plausibility by artificial intelligence inside or outside at least one vehicle. Especially in the case of artificial intelligence outside the vehicle, the planned trajectory of the autonomous vehicle can be proactively transmitted to the cloud in real time through vehicle-to-X communication. On this cloud, the plausibility of the vehicle trajectory can be verified using the data of the high-precision map and on the basis of the current speed, the direction of travel and the vehicle dynamics of the vehicle.

The planned trajectory can preferably be checked in the simulation by existing or additional artificial intelligence which executes the desired trajectory by means of the simulation model and subsequently compares the position resulting from the simulation with the map and the other traffic participants on the map. In such an embodiment, the vehicle may be placed in a safe state from the cloud if at least one environmental sensor has degraded performance. It is possible to check within the cloud which of these sensors is defective by transmitting the sensor data or the calculated trajectory for the artificial intelligence of the autonomous driving to the cloud and evaluating it there within a simulation. In this case, the vehicle environment sensors are again successively deactivated step by step and the reaction to the planned trajectory is checked within the simulation of the artificial intelligence. In the direct transfer of sensor data, trajectories are planned or calculated within the artificial intelligence on the cloud, wherein the results can be validated for plausibility within the artificial intelligence hosting the simulation. This is particularly relevant if it is intended to detect whether a fault is present in the artificial intelligence of the vehicle and whether all ambient sensors are also still operating without any doubt. When transmitting a trajectory calculated by artificial intelligence of a vehicle, it is also possible to perform plausibility verification only on the trajectory itself within the cloud.

According to a further embodiment, the sensor data within a defined time period is intermediately stored in a memory unit and subsequently analyzed by artificial intelligence for the purpose of performing plausibility verification. In this case, the data of the vehicle surroundings sensors can be intermediately stored in the vehicle memory for a specific period of time, for example 5 minutes. After the vehicle has been brought into a safe state, the sensor data from the vehicle memory are retransmitted to the artificial intelligence for autonomous driving and the generated trajectory is simulated for plausibility verification by means of the artificial intelligence for simulating the trajectory. In this case, the respective artificial intelligence is always fed with different sensor data, wherein at least one environmental sensor is always deactivated and the results are calculated by the artificial intelligence hosting the simulation and compared in a plurality of steps for all environmental sensors. In this way, environmental sensors with poor performance can be identified in a simple manner.

The trace that was trusted during the time that just one determined environmental sensor was deactivated in these calculations after considering the stored data provides the following cues: which environmental sensor is defective or degrades in its performance. As a check, the respective environmental sensor can be reincorporated into the calculation and the result can be compared with the calculation if the environmental sensor is activated. If, for example, a sudden change in position results, this also indicates that a correct environmental sensor with poor performance is identified, which can now be deactivated continuously in a subsequent step. Alternatively or additionally, a warning or fault report is generated. The embodiment can alternatively be carried out in the cloud by analyzing the respective sensor data that has been intermediately stored in the vehicle by means of the method of the embodiment in the cloud. For this purpose, these sensor data can be transmitted to the server unit or the cloud beforehand.

According to another embodiment, the sensor data is analyzed by artificial intelligence in parallel with the automated driving mode implemented. Even during normal driving, the sensor data is analyzed by background calculations in parallel paths for artificial intelligence of autonomous driving. This is done in a similar manner, by always excluding certain environmental sensors when calculating the trajectory and by means of artificial intelligence governing the simulation, simulating the generated trajectory in the controller. In this way, environmental sensors with poor performance can be identified and deactivated in almost real time.

The following advantages are obtained by the method of the invention:

safety in road traffic is significantly improved by identifying environmental sensors that do not have faults, but nevertheless provide poor performance for automated driving.

By means of the artificial intelligence in the vehicle or in the cloud, which governs the simulation, the planned trajectory for the determined vehicle variant can already be compared with the target trajectory or the desired behavior before execution by the vehicle.

Purposeful sequential switching off of the environmental sensors leads to a simple identification of environmental sensors with degraded performance.

By storing sensor data of the environmental sensors in the vehicle memory within a specific time, a trajectory can be repeatedly calculated for a plurality of activated or deactivated environmental sensors by simulation for the purpose of identifying a fault.

Drawings

In the following, preferred embodiments of the invention are explained in detail on the basis of strongly simplified schematic drawings. Shown here are:

figure 1 is a schematic diagram of a system with a vehicle and a server unit outside the vehicle,

fig. 2 is a schematic flow chart for visually illustrating a method according to an embodiment of the present invention.

Detailed Description

Fig. 1 shows a schematic view of a system 1 with a vehicle 2 and a server unit 4 outside the vehicle.

The vehicle 2 has a plurality of sensors 6, 8. According to this embodiment, the two sensors 6,8 are environmental sensors, wherein the sensor 6 is a radar sensor and the second sensor 8 is a GNSS sensor.

The sensors 6,8 are connected to a controller 10 in a data-transmitting manner. Artificial intelligence is set up on the control unit 10, which can calculate, simulate and analyze the trajectory on the basis of the sensor data of the sensors 6, 8.

Further, the controller 10 is connected to a vehicle-side memory unit 12. The sensor data can thus be intermediately stored in the memory unit 12 for a defined period of time.

The vehicle 2 has a communication unit 14. The controller 10 can establish a wireless communication connection 16 with the server unit 4 outside the vehicle by means of the communication unit 14 and transmit data. Here, the communication connection 16 is a vehicle-to-X communication connection 16 based on a mobile radio standard.

The server unit 4 is configured as a cloud and has at least one memory 18. At least one artificial intelligence may also be stored on the accessor 18 and may be executed by the server unit 4. In particular, a connection for exchanging data and sending control commands can be established between the server unit 4 and the controller 10.

A schematic flow chart for visually illustrating a method 20 according to an embodiment of the present invention is shown in fig. 2.

The method may be performed in particular by the system 1, the controller 10 or the external server unit 4.

In step 21, the sensor data of the sensors 6,8 are read by the controller 10. These sensor data can be stored in the memory unit 12 or forwarded to the server unit 4 via the communication connection 16, if necessary.

In step 22, the nominal trajectory of the vehicle 3 is calculated by the controller 10.

The target trajectory can then be traveled 23 by the vehicle 2. Alternatively, the nominal trajectory can be simulated 24 by artificial intelligence based on the sensor data. The simulation may be performed based on the sensor data. Here, the artificial intelligence may be stored on the controller 10 or the server unit 4.

In a further step 25, the actual or simulated trajectory traveled by vehicle 2 is compared with the target trajectory.

In the event that a deviation of the actual or simulated trajectory from the setpoint trajectory is determined 26, the performance of at least one sensor 6,8 is checked by the controller 10.

The check is carried out 27 in such a way that each sensor 6,8 of the vehicle 2 is deactivated in an alternating manner and the target trajectory is followed by the vehicle 2 by means of at least one selected sensor 6,8 or is simulated by artificial intelligence. Thereby identifying a sensor with lower performance.