Adaptive prediction ensemble for motion forecasting

A system, comprising:

• circuitry configured to:
• acquire input data including road map images and historical trajectory information of a set of agents in the road map images;
• transform the input data into a vectorized representation;
• generate a first candidate trajectory prediction for the set of agents by application of a motion prediction neural network on the vectorized representation;
• generate a second candidate trajectory prediction for the set of agents by application of a rule-based prediction model on the acquired input data;
• generate ranking results for the first candidate trajectory prediction and the second candidate trajectory prediction based on a routing function network;
• select a trajectory for at least one agent based on the ranking results;
• and
• a vehicle control system configured to control motion of an ego agent based on the trajectory selected by the circuitry.

A method, comprising:

• acquiring input data including road map images and historical trajectory information of a set of agents in the road map images;
• transforming the input data into a vectorized representation;
• generating a first candidate trajectory prediction for the set of agents by application of a motion prediction neural network on the vectorized representation;
• generating a second candidate trajectory prediction for the set of agents by application of a rule-based prediction model on the acquired input data;
• generating ranking results for the first candidate trajectory prediction and the second candidate trajectory prediction based on a routing function network;
• selecting a trajectory for at least one agent based on the ranking results;
• and
• controlling motion of an ego agent based on the selected trajectory.