| CPC B60W 30/143 (2013.01) [B60W 40/105 (2013.01); B60W 50/082 (2013.01); B60W 50/085 (2013.01); B60W 60/0011 (2020.02); B60W 2520/10 (2013.01); B60W 2720/24 (2013.01)] | 24 Claims |
|
1. A computer-implemented method, comprising:
obtaining, by a planning circuit on a vehicle, a proposed trajectory for the vehicle in response to a driving scenario;
providing, by the planning circuit, the proposed trajectory for the vehicle as an input to a control circuit on the vehicle that includes a machine learning network for generating one or more estimated speed profiles, the machine learning network trained using training vectors corresponding to a plurality of trajectories and a plurality of corresponding speed profiles, wherein the plurality of trajectories and the plurality of corresponding speed profiles are determined based on past driving behavior in different driving scenarios;
obtaining, by the planning circuit, an estimated speed profile of the vehicle, and a confidence score, as an output of the control circuit, wherein the confidence score represents a similarity of the estimated speed profile of the vehicle to an actual speed profile of the vehicle that would be generated by the control circuit for the proposed trajectory for the vehicle;
determining whether the confidence score meets a confidence threshold; and
in accordance with a determination that the confidence score exceeds the confidence threshold, operating, by the control circuit on the vehicle, the vehicle along the proposed trajectory.
|