CVROJul 5, 2025

Breaking Imitation Bottlenecks: Reinforced Diffusion Powers Diverse Trajectory Generation

arXiv:2507.04049v217 citationsh-index: 12
Originality Highly original
AI Analysis

This work addresses the mode collapse problem in imitation learning for autonomous driving, offering a solution to enhance generalization in complex real-world scenarios.

The paper tackles the problem of conservative and homogeneous behaviors in end-to-end autonomous driving by proposing DIVER, a framework that integrates reinforcement learning with diffusion-based generation to produce diverse and feasible trajectories, significantly improving trajectory diversity on benchmarks like NAVSIM, Bench2Drive, and nuScenes.

Most end-to-end autonomous driving methods rely on imitation learning from single expert demonstrations, often leading to conservative and homogeneous behaviors that limit generalization in complex real-world scenarios. In this work, we propose DIVER, an end-to-end driving framework that integrates reinforcement learning with diffusion-based generation to produce diverse and feasible trajectories. At the core of DIVER lies a reinforced diffusion-based generation mechanism. First, the model conditions on map elements and surrounding agents to generate multiple reference trajectories from a single ground-truth trajectory, alleviating the limitations of imitation learning that arise from relying solely on single expert demonstrations. Second, reinforcement learning is employed to guide the diffusion process, where reward-based supervision enforces safety and diversity constraints on the generated trajectories, thereby enhancing their practicality and generalization capability. Furthermore, to address the limitations of L2-based open-loop metrics in capturing trajectory diversity, we propose a novel Diversity metric to evaluate the diversity of multi-mode predictions.Extensive experiments on the closed-loop NAVSIM and Bench2Drive benchmarks, as well as the open-loop nuScenes dataset, demonstrate that DIVER significantly improves trajectory diversity, effectively addressing the mode collapse problem inherent in imitation learning.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes