Display of compression environment features for vehicle behavior prediction

摘要

A method for indicating an ambient environment of an autonomous ADV (ADV). The method indicates a surrounding environment using a first set of features from a high definition map and a second set of features from a target object in the surrounding environment. The first feature set is extracted from the high-definition map by a convolutional neural network, and the second feature set is handcrafted features from the target object during a predetermined number of autonomous ADV past travel cycles. .. The first feature set and the second feature set are connected in series and provided to multiple fully connected layers of the CNN to predict the behavior of the target object. In one embodiment, the operations in the method may be repeated for each autonomous ADV driving cycle. [Selection diagram] Fig. 6