Evaluating driving data using autonomous vehicle control system

A system comprising:

• at least one processor;
• and
• memory storing instructions that, when executed, cause the at least one processor to be operable to:
• process a current trajectory of a vehicle for an initial instance of manual driving data to generate a predicted subsequent trajectory of the vehicle that is predicted to follow the current trajectory and for a subsequent instance of manual driving data;
• determine driving latency based on the initial instance of the manual driving data, the subsequent instance of the manual driving data, and the predicted subsequent trajectory of the vehicle that is predicted to follow the current trajectory and for the subsequent instance of manual driving data, wherein the driving latency measures a difference in a human reacting to an environment of the vehicle compared to an autonomous vehicle control system
• determine, based on comparing the predicted subsequent trajectory of the vehicle that is predicted to follow the current trajectory and for the subsequent instance of manual driving data with an actual subsequent trajectory of the vehicle that actually followed the current trajectory and for the subsequent instance of manual driving data, and based on the driving latency, a difference measure;
• and
• determine, based on the difference measure, whether to utilize one or more particular portions of the manual driving data, that are based on the initial instance of the manual driving data and/or the subsequent instance of the manual driving data, in training of the autonomous vehicle control system;
• and
• in response to determining to utilize the one or more particular portions of the manual driving data in training of the autonomous vehicle control system:
• cause the one or more particular portions of the manual driving data to be included for subsequent use in training the autonomous vehicle control system.

A non-transitory computer-readable storage medium storing instructions that, when executed by at least one processor, cause the at least one processor perform operations to:

• process a current trajectory of a vehicle for an initial instance of manual driving data to generate a predicted subsequent trajectory of the vehicle that is predicted to follow the current trajectory and for a subsequent instance of manual driving data;
• determine driving latency based on the initial instance of the manual driving data, the subsequent instance of the manual driving data, and the predicted subsequent trajectory of the vehicle that is predicted to follow the current trajectory and for the subsequent instance of manual driving data, wherein the driving latency measures a difference in a human reacting to an environment of the vehicle compared to an autonomous vehicle control system;
• determine, based on comparing the predicted subsequent trajectory of the vehicle that is predicted to follow the current trajectory and for the subsequent instance of manual driving data with an actual subsequent trajectory of the vehicle that actually followed the current trajectory and for the subsequent instance of manual driving data, and based on the driving latency, a difference measure;
• and
• determine, based on the difference measure, whether to utilize one or more particular portions of the manual driving data, that are based on the initial instance of the manual driving data and/or the subsequent instance of the manual driving data, in training of the autonomous vehicle control system;
• and
• in response to determining to utilize the one or more particular portions of the manual driving data in training of the autonomous vehicle control system:
• cause the one or more particular portions of the manual driving data to be included for subsequent use in training the autonomous vehicle control system.