| CPC B60W 30/0956 (2013.01) [B60W 50/10 (2013.01); B60W 50/14 (2013.01); G06N 3/04 (2013.01); G06T 7/38 (2017.01); B60W 2420/403 (2013.01); B60W 2554/20 (2020.02); B60W 2554/404 (2020.02)] | 20 Claims |
|
1. A computer-implemented method for completing risk object identification comprising:
receiving image data associated with a monocular image of a surrounding environment of an ego vehicle;
analyzing the image data and completing semantic waypoint labeling, of at least one region of the surrounding environment of the ego vehicle, to determine features of the surrounding environment and a path projection that corresponds to a travel path of the ego vehicle;
receiving one or more object profiles associated with at least one of a dynamic object and a static object based on the features of the surrounding environment;
completing counterfactual scenario augmentation, while the ego vehicle is travelling on the travel path and using a neural network, with respect to the at least one region by inpainting the at least one of the dynamic object and the static object at pixels within the travel path of the ego vehicle in the monocular image based on the one or more object profiles; and
determining, in real time while the ego vehicle is travelling on the travel path, at least one driver intention and at least one driver response by the neural network augmented in real time with the counterfactual scenario augmentation of the at least one region,
wherein at least one vehicle system is controlled based on at least one of: the at least one driver intention and the at least one driver response, and
wherein the inpainted object is generated from a stored object associated with the one or more object profiles.
|