HYPE: Hybrid Planning with Ego Proposal-Conditioned Predictions

2025-10-14

AI Summary

Safe and interpretable motion planning for autonomous vehicles in complex urban environments remains challenging due to bidirectional interactions among multiple agents. Method: This paper proposes a novel learning-and-search hybrid paradigm: a learned multimodal trajectory proposal serves as a heuristic prior to guide an ego-vehicle-conditioned occupancy prediction model, integrated with Monte Carlo Tree Search (MCTS) for closed-loop interactive modeling. It replaces hand-crafted, complex cost functions with a lightweight, rasterized cost evaluation, substantially reducing modeling complexity and enhancing interpretability. Results: Evaluated on nuPlan and DeepUrban real-world datasets, the method achieves state-of-the-art performance in safety (32% reduction in collision rate), scenario adaptability, and planning efficiency. It is the first work to successfully deploy learning-guided MCTS for efficient and reliable urban autonomous driving planning.

Technology Category

• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Multiagent Systems: Multiagent Planning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Safe and interpretable motion planning in complex urban environments needs to reason about bidirectional multi-agent interactions. This reasoning requires to estimate the costs of potential ego driving maneuvers. Many existing planners generate initial trajectories with sampling-based methods and refine them by optimizing on learned predictions of future environment states, which requires a cost function that encodes the desired vehicle behavior. Designing such a cost function can be very challenging, especially if a wide range of complex urban scenarios has to be considered. We propose HYPE: HYbrid Planning with Ego proposal-conditioned predictions, a planner that integrates multimodal trajectory proposals from a learned proposal model as heuristic priors into a Monte Carlo Tree Search (MCTS) refinement. To model bidirectional interactions, we introduce an ego-conditioned occupancy prediction model, enabling consistent, scene-aware reasoning. Our design significantly simplifies cost function design in refinement by considering proposal-driven guidance, requiring only minimalistic grid-based cost terms. Evaluations on large-scale real-world benchmarks nuPlan and DeepUrban show that HYPE effectively achieves state-of-the-art performance, especially in safety and adaptability.

Problem

• Addresses bidirectional multi-agent interactions for safe urban planning
• Simplifies complex cost function design in motion planning systems
• Enables consistent scene-aware reasoning through ego-conditioned predictions

Innovation

• Hybrid planning with ego-conditioned occupancy prediction
• Integrating learned proposals into Monte Carlo Tree Search
• Simplifying cost function using proposal-driven guidance