Autonomous vehicle fleet model training and testing
First Claim
1. A vehicle comprising:
- a vehicle controller operably connected to at least one sensor, a network interface, and a drive system to control operation of the vehicle, the vehicle controller being configured to;
receive an indication that the vehicle controller has unallocated computational resources;
receive a command to utilize the unallocated computational resources, the command causing the vehicle controller to be further configured to conduct one or more of;
testing an experimental machine learning (ML) model based at least in part on causing the drive system to control operation of the vehicle using the experimental ML model, the testing yielding a test result, ortraining a target ML model to create a trained ML model, based at least in part on causing the drive system to control operation of the vehicle using the target ML model.
1 Assignment
0 Petitions
Accused Products
Abstract
A method and system of using excess computational resources on autonomous vehicles. Such excess computational resources may be available during periods of low demand, or other periods of idleness (e.g., parking). Where portions of computing resources are available amongst a fleet of vehicles, such excess computing resources may be pooled as a single resource. The excess computational resources may be used, for example, to train and/or test machine-learning models. Performance metrics of such models may be determined using hardware and software on the autonomous vehicle, for example sensors. Models having performance metrics outperforming current models may be considered as validated models. Validated models may be transmitted to a remote computing system for dissemination to a fleet of vehicles.
71 Citations
20 Claims
-
1. A vehicle comprising:
a vehicle controller operably connected to at least one sensor, a network interface, and a drive system to control operation of the vehicle, the vehicle controller being configured to; receive an indication that the vehicle controller has unallocated computational resources; receive a command to utilize the unallocated computational resources, the command causing the vehicle controller to be further configured to conduct one or more of; testing an experimental machine learning (ML) model based at least in part on causing the drive system to control operation of the vehicle using the experimental ML model, the testing yielding a test result, or training a target ML model to create a trained ML model, based at least in part on causing the drive system to control operation of the vehicle using the target ML model. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
16. A method of controlling a vehicle comprising:
-
receiving an indication that a vehicle controller of the vehicle has or will have unallocated computational resources at a time period; based at least in part on receiving the indication, causing the vehicle controller to one or more of; train a target machine-learning model, creating a trained model;
ortest an experimental model; and control the vehicle according to an output of the trained model or the experimental model. - View Dependent Claims (17, 18)
-
-
19. A non-transitory computer-readable medium having a set of instructions that, when executed, cause one or more processors to perform operations comprising:
-
determining that a vehicle has unallocated computational resources during a time period; transmitting to the vehicle instructions to conduct a test mission during the time period and one or more of an experimental model or instructions to train a target model stored at the vehicle; receiving one or more of sensor data, an updated target model, or performance metrics from the vehicle that correspond to completion of at least a portion of the test mission; and determining that one or more of the updated target model or the experimental model changed performance of the vehicle. - View Dependent Claims (20)
-
Specification