System learning update for online sensor alignment
A system and method for system learning updates for online sensor alignment includes receiving sensor data detected by a sensor system of a vehicle, generating, using a coordinate transformation matrix (CTM), a sensor alignment result, and determining, based on a degradation detection model, that a deviation between the sensor alignment result and one or more of a fleet model, a vehicle model, and a system model exceeds an onboard degradation threshold. The system and method also includes determining, based on an offline degradation model, that a deviation between the sensor alignment result and an offline model result exceeds an offline degradation threshold, and triggering corner case data collection to collect additional sensor data detected by the sensor system of the vehicle.
1 . A computer-implemented method when executed on data processing hardware causes the data processing hardware to perform operations comprising:
receiving sensor data detected by a sensor system of a vehicle;
generating, using a coordinate transformation matrix (CTM), a sensor alignment result;
determining, based on a degradation detection model, that a deviation between the sensor alignment result and one or more of a fleet model, a vehicle model, and a system model exceeds an onboard degradation threshold;
determining, based on an offline degradation model, that a deviation between the sensor alignment result and an offline model result exceeds an offline degradation threshold; and
triggering corner case data collection to collect additional sensor data detected by the sensor system of the vehicle.
2 . The method of claim 1 , wherein the operations further comprise:
generating, based on the additional sensor data, a corrected CTM; and
communicating the corrected CTM to the vehicle.
3 . The method of claim 2 , wherein generating the corrected CTM and the updated sensor data collection trigger is based on an offline alignment model trained to generate the corrected CTM based on the sensor data and the additional sensor data.
4 . The method of claim 1 , wherein the fleet model comprises a statistical average fleet sensor alignment based on the sensor data.
5 . The method of claim 1 , wherein the vehicle model comprises a statistical average model sensor alignment of vehicle models similar to a model of the vehicle based on the sensor data.
6 . The method of claim 1 , wherein the system model comprises a statistical average sensor alignment of the sensor system of the vehicle based on the sensor data.
7 . The method of claim 1 , wherein determining, based on the offline degradation model, that the deviation between the sensor alignment result and the offline model result exceeds the offline degradation threshold comprises:
generating, using the offline degradation model, the offline model result;
providing, as input to a performance model, the sensor alignment result and the offline model result; and
receiving, as output from the performance model, an indication that the deviation between the sensor alignment result and the offline model result exceeds the offline degradation threshold.
8 . The method of claim 7 , wherein the performance model comprises a machine learning model.
9 . The method of claim 7 , wherein the performance model comprises a rule-based model.
10 . The method of claim 1 , wherein triggering the corner case data collection to collect the additional sensor data comprises collecting a context of the vehicle.
11 . A system comprising:
data processing hardware; and
memory hardware in communication with the data processing hardware, the memory hardware storing instructions that when executed on the data processing hardware cause the data processing hardware to perform operations comprising:
receiving sensor data detected by a sensor system of a vehicle;
generating, using a coordinate transformation matrix (CTM), a sensor alignment result;
determining, based on a degradation detection model, that a deviation between the sensor alignment result and one or more of a fleet model, a vehicle model, and a system model exceeds an onboard degradation threshold;
determining, based on an offline degradation model, that a deviation between the sensor alignment result and an offline model result exceeds an offline degradation threshold; and
triggering corner case data collection to collect additional sensor data detected by the sensor system of the vehicle.
12 . The system of claim 11 , wherein the operations further comprise:
generating, based on the additional sensor data, a corrected CTM; and
communicating the corrected CTM to the vehicle.
13 . The system of claim 12 , wherein generating the corrected CTM and the updated sensor data collection trigger is based on an offline alignment model trained to generate the corrected CTM based on the sensor data and the additional sensor data.
14 . The system of claim 11 , wherein the fleet model comprises a statistical average fleet sensor alignment based on the sensor data.
15 . The system of claim 11 , wherein the vehicle model comprises a statistical average model sensor alignment of vehicle models similar to a model of the vehicle based on the sensor data.
16 . The system of claim 11 , wherein the system model comprises a statistical average sensor alignment of the sensor system of the vehicle based on the sensor data.
17 . The system of claim 11 , wherein determining, based on the offline degradation model, that the deviation between the sensor alignment result and the offline model result exceeds the offline degradation threshold comprises:
generating, using the offline degradation model, the offline model result;
providing, as input to a performance model, the sensor alignment result and the offline model result; and
receiving, as output from the performance model, an indication that the deviation between the sensor alignment result and the offline model result exceeds the offline degradation threshold.
18 . The system of claim 17 , wherein the performance model comprises a machine learning model.
19 . The system of claim 17 , wherein the performance model comprises a rule-based model.
20 . The system of claim 11 , wherein triggering the corner case data collection to collect the additional sensor data comprises collecting a context of the vehicle.