Self-diagnosis of faults in an autonomous driving system
This application discloses a computing system to self-diagnosis of faults for an assisted or automated driving system of a vehicle. The computing system can populate measurement data collected by sensors mounted in the vehicle into an environmental model, and detect an object proximate to the vehicle based on the measurement data in the environmental model. The computing system can identify a fault in at least one of the sensors by performing one or more safety cross-checks including one or more of comparing the environmental modal against one or more data models populated with sensor measurements or objects detected by different sensors, analyzing estimated motion of the vehicle or the object for aberrant movement, identifying missing measurement data, identifying divergent object classifications, or identifying operational sensor characteristics. A control system for the vehicle can configured to control operation of the vehicle based, at least in part, on the identified fault.
1. A method comprising:
populating, by a computing system, measurement data collected by a sensor system having sensors mounted in a vehicle into an environmental model associated with the vehicle;
detecting, by the computing system, an object proximate to the vehicle based on the measurement data populated in the environmental model by fusing at least a portion of the measurement data into a detection event corresponding to the object;
identifying, by the computing system, a fault in at least one of the sensors in the sensor system based, at least in part, on the detection of the object proximate to the vehicle by identifying the measurement data from the at least one of the sensors that was not fused in the detection event even though the at least one of the sensors had visibility to a location associated with the detection event; and
recalibrating, by the computing system, at least a portion of the sensor system based, at least in part, on the identified fault in the at least one of the sensors.
2. The method of claim 1 , further comprising generating, by the computing system, a fault message based, at least in part, on identification of the fault in the at least one of the sensors, wherein the fault message is configured to identify the measurement data collected by the sensors associated with the fault.
3. The method of claim 1 , wherein identifying the fault in the at least one of the sensors further comprises:
comparing the environmental model populated with the measurement data against one or more data models received from a source external to the vehicle, wherein the data models are populated with sensor measurements or objects detected by different sensors; and
detecting the fault in the at least one of the sensors based on differences between the environmental model and the one or more data models.
4. The method of claim 1 , wherein detecting the object proximate to the vehicle motion further comprises estimating motion of the vehicle or the object based on the measurement data, and identifying the fault in the at least one of the sensors when the estimated motion of the vehicle or the object exceeds a threshold level of movement.
5. The method of claim 1 , further comprising comparing, by the computing system, operational characteristics of the sensor to at least one of an electrical threshold or an environmental threshold, wherein identifying the fault in the at least one of the sensors when the comparison indicates the sensor has operational characteristics that exceed the electrical threshold or the environmental threshold.
6. The method of claim 1 , wherein detecting the object proximate to the vehicle motion further comprises utilizing multiple classifiers to generate classifications associated with the object, and further comprising identifying, by the computing system, a fault associated with at least one of the classifiers based on a comparison of the classifications associated with the object.
7. An apparatus comprising at least one memory device storing instructions configured to cause one or more processing devices to perform operations comprising:
populating measurement data collected by a sensor system having sensors mounted in a vehicle into an environmental model associated with the vehicle;
detecting an object proximate to the vehicle based on the measurement data populated in the environmental model by fusing at least a portion of the measurement data into a detection event corresponding to the object; and
identifying a fault in at least one of the sensors in the sensor system based, at least in part, on the detection of the object proximate to the vehicle by identifying the measurement data from the at least one of the sensors that was not fused in the detection event even though the at least one of the sensors had visibility to a location associated with the detection event; and
recalibrating at least a portion of the sensor system based, at least in part, on the identified fault in the at least one of the sensors.
8. The apparatus of claim 7 , wherein the instructions are further configured to cause the one or more processing devices to perform operations comprising generating a fault message based, at least in part, on identification of the fault in the at least one of the sensors, wherein the fault message is configured to identify the measurement data collected by the sensors associated with the fault.
9. The apparatus of claim 7 , wherein the instructions are further configured to cause the one or more processing devices to perform operations comprising:
comparing the environmental model populated with the measurement data against one or more data models received from a source external to the vehicle, wherein the data models are populated with sensor measurements or objects detected by different sensors; and
detecting the fault in the at least one of the sensors based on differences between the environmental model and the one or more data models.
10. The apparatus of claim 7 , wherein the instructions are further configured to cause the one or more processing devices to perform operations comprising:
estimating motion of the vehicle or the object based on the measurement data; and
identifying the fault in the at least one of the sensors when the estimated motion of the vehicle or the object exceeds a threshold level of movement.
11. The apparatus of claim 7 , wherein the instructions are further configured to cause the one or more processing devices to perform operations comprising:
comparing operational characteristics of the sensor to at least one of an electrical threshold or an environmental threshold; and
identifying the fault in the at least one of the sensors when the comparison indicates the sensor has operational characteristics that exceed the electrical threshold or the environmental threshold.
12. The apparatus of claim 7 , wherein the instructions are further configured to cause the one or more processing devices to perform operations comprising:
utilizing multiple classifiers to generate classifications associated with the object; and
identifying a fault associated with at least one of the classifiers based on a comparison of the classifications associated with the object.
13. A system comprising:
a memory device configured to store machine-readable instructions; and
a computing system including one or more processing devices, in response to executing the machine-readable instructions, configured to:
populate measurement data collected by a sensor system having sensors mounted in a vehicle into an environmental model associated with the vehicle;
detect an object proximate to the vehicle based on the measurement data populated in the environmental model by fusing at least a portion of the measurement data into a detection event corresponding to the object; and
identify a fault in at least one of the sensors in the sensor system based, at least in part, on the detection of the object proximate to the vehicle by identifying the measurement data from the at least one of the sensors that was not fused in the detection event even though the at least one of the sensors had visibility to a location associated with the detection event; and
recalibrate at least a portion of the sensor system based, at least in part, on the identified fault in the at least one of the sensors.
14. The system of claim 13 , wherein the one or more processing devices, in response to executing the machine-readable instructions, are configured to generate a fault message based, at least in part, on identification of the fault in the at least one of the sensors, and wherein the fault message is configured to identify the measurement data collected by the sensors associated with the fault.
15. The system of claim 13 , wherein the one or more processing devices, in response to executing the machine-readable instructions, are configured to:
compare the environmental model populated with the measurement data against one or more data models received from a source external to the vehicle, wherein the data models are populated with sensor measurements or objects detected by different sensors; and
detect the fault in the at least one of the sensors based on differences between the environmental model and the one or more data models.
16. The system of claim 13 , wherein the one or more processing devices, in response to executing the machine-readable instructions, are configured to:
estimate motion of the vehicle or the object based on the measurement data; and
identify the fault in the at least one of the sensors when the estimated motion of the vehicle or the object exceeds a threshold level of movement.
17. The system of claim 13 , wherein the one or more processing devices, in response to executing the machine-readable instructions, are configured to:
utilize multiple classifiers to generate classifications associated with the object; and
identify a fault associated with at least one of the classifiers based on a comparison of the classifications associated with the object.