Tractable Joint Prediction and Planning over Discrete Behavior Modes for Urban Driving
This addresses the challenge of reactive planning in dynamic urban driving for autonomous vehicles, representing an incremental improvement over existing methods.
The paper tackles the problem of integrating multimodal trajectory forecasting models with downstream planners for autonomous driving by using learned anchor embeddings to parameterize discrete behavior modes, enabling tractable closed-loop planning that avoids the frozen robot problem and outperforms previous state-of-the-art in CARLA on dense traffic scenarios at realistic speeds.
Significant progress has been made in training multimodal trajectory forecasting models for autonomous driving. However, effectively integrating these models with downstream planners and model-based control approaches is still an open problem. Although these models have conventionally been evaluated for open-loop prediction, we show that they can be used to parameterize autoregressive closed-loop models without retraining. We consider recent trajectory prediction approaches which leverage learned anchor embeddings to predict multiple trajectories, finding that these anchor embeddings can parameterize discrete and distinct modes representing high-level driving behaviors. We propose to perform fully reactive closed-loop planning over these discrete latent modes, allowing us to tractably model the causal interactions between agents at each step. We validate our approach on a suite of more dynamic merging scenarios, finding that our approach avoids the $\textit{frozen robot problem}$ which is pervasive in conventional planners. Our approach also outperforms the previous state-of-the-art in CARLA on challenging dense traffic scenarios when evaluated at realistic speeds.