Conditional Prediction by Simulation for Automated Driving
This work addresses the challenge of cooperative planning for automated driving, which is incremental as it builds on existing modular systems by integrating prediction and planning more closely.
The paper tackles the problem of enabling cooperative maneuvers in modular automated driving systems by introducing a prediction model that generates conditional dependencies between trajectories using microscopic traffic simulation and a behavior model trained via Adversarial Inverse Reinforcement Learning, with results demonstrated through example scenarios.
Modular automated driving systems commonly handle prediction and planning as sequential, separate tasks, thereby prohibiting cooperative maneuvers. To enable cooperative planning, this work introduces a prediction model that models the conditional dependencies between trajectories. For this, predictions are generated by a microscopic traffic simulation, with the individual traffic participants being controlled by a realistic behavior model trained via Adversarial Inverse Reinforcement Learning. By assuming various candidate trajectories for the automated vehicle, we generate predictions conditioned on each of them. Furthermore, our approach allows the candidate trajectories to adapt dynamically during the prediction rollout. Several example scenarios are available at https://conditionalpredictionbysimulation.github.io/.