Conditional Prediction by Simulation for Automated Driving

TL;DR

This paper introduces a conditional prediction model for automated driving that uses microscopic traffic simulation and adversarial learning to enable cooperative maneuvers and dynamic trajectory adaptation.

Contribution

It presents a novel prediction approach modeling conditional dependencies between trajectories using simulation and learned behavior models.

Findings

Enables cooperative planning in automated driving.

Allows dynamic adaptation of candidate trajectories.

Demonstrates realistic traffic scenario predictions.

Abstract

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/.