TECHNOLOGY FOR THE APPLICATION OF DRIVING STANDARDS FOR THE BEHAVIOR PREDICTION OF AUTOMATED VEHICLES

摘要

Systems, devices, and methods can provide technology that generates a series of time-stamped object graphs based on object trajectory histories derived from external object data for multiple external objects, such as vehicles. The technology can also generate a set of relational object representations based on the set of time-stamped object graphs via a first neural network, such as a graph attention network, and predicted object trajectories for the determine multiple external objects based on the set of relational object representations. The technology can also modify an autonomous vehicle's behavior based on the predicted object trajectories and real-time perception error information.