US 11,897,490 B2
Autonomous driving vehicle health monitoring
Kai Chen, San Jose, CA (US); and Pingfan Meng, Fremont, CA (US)
Assigned to Pony AI Inc., Grand Cayman (KY)
Filed by Pony AI Inc., Grand Cayman (KY)
Filed on Jan. 31, 2023, as Appl. No. 18/162,169.
Application 18/162,169 is a continuation of application No. 17/073,980, filed on Oct. 19, 2020, granted, now 11,565,708.
Prior Publication US 2023/0182753 A1, Jun. 15, 2023
Int. Cl. B60W 50/029 (2012.01); B60W 50/02 (2012.01); B60W 50/038 (2012.01)
CPC B60W 50/029 (2013.01) [B60W 50/0205 (2013.01); B60W 50/038 (2013.01); B60W 2050/021 (2013.01); B60W 2050/0297 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for vehicle component fault detection, the method comprising:
monitoring a health status of a first vehicle component over a period of time;
detecting failure of the first vehicle component;
determining a failure level associated with the failure of the first vehicle component;
determining, based at least in part on the failure level and a component type of the first vehicle component, a vehicle component task reassignment, the component type of the first vehicle component comprising a graphics processing unit (GPU) or a field-programmable gate array (FPGA); and
implementing the vehicle component task reassignment at least in part by offloading at least a portion of the task processing performed by the first vehicle component, prior to failure of the first vehicle component, to a second vehicle component, wherein an operational health of the second vehicle component satisfies one or more operational health criteria, and wherein the implementing of the vehicle component task reassignment comprises:
selecting an available candidate reassignment that offloads the portion of the task processing to a highest number of GPUs and a lowest number of other components over another candidate reassignment that offloads the portion of the task processing to a lower number of GPUs and a higher number of the other components, while satisfying operational health criteria of the one or more other GPUs and operation rules of the vehicle; and
selectively implementing the vehicle component task reassignment based on the available candidate reassignment.