RealDrive: Retrieval-Augmented Driving with Diffusion Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ding, Wenhao, Veer, Sushant, Chen, Yuxiao, Cao, Yulong, Xiao, Chaowei, Pavone, Marco
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915313972412416
author Ding, Wenhao
Veer, Sushant
Chen, Yuxiao
Cao, Yulong
Xiao, Chaowei
Pavone, Marco
author_facet Ding, Wenhao
Veer, Sushant
Chen, Yuxiao
Cao, Yulong
Xiao, Chaowei
Pavone, Marco
contents Learning-based planners generate natural human-like driving behaviors by learning to reason about nuanced interactions from data, overcoming the rigid behaviors that arise from rule-based planners. Nonetheless, data-driven approaches often struggle with rare, safety-critical scenarios and offer limited controllability over the generated trajectories. To address these challenges, we propose RealDrive, a Retrieval-Augmented Generation (RAG) framework that initializes a diffusion-based planning policy by retrieving the most relevant expert demonstrations from the training dataset. By interpolating between current observations and retrieved examples through a denoising process, our approach enables fine-grained control and safe behavior across diverse scenarios, leveraging the strong prior provided by the retrieved scenario. Another key insight we produce is that a task-relevant retrieval model trained with planning-based objectives results in superior planning performance in our framework compared to a task-agnostic retriever. Experimental results demonstrate improved generalization to long-tail events and enhanced trajectory diversity compared to standard learning-based planners -- we observe a 40% reduction in collision rate on the Waymo Open Motion dataset with RAG.
format Preprint
id arxiv_https___arxiv_org_abs_2505_24808
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RealDrive: Retrieval-Augmented Driving with Diffusion Models
Ding, Wenhao
Veer, Sushant
Chen, Yuxiao
Cao, Yulong
Xiao, Chaowei
Pavone, Marco
Robotics
Artificial Intelligence
Learning-based planners generate natural human-like driving behaviors by learning to reason about nuanced interactions from data, overcoming the rigid behaviors that arise from rule-based planners. Nonetheless, data-driven approaches often struggle with rare, safety-critical scenarios and offer limited controllability over the generated trajectories. To address these challenges, we propose RealDrive, a Retrieval-Augmented Generation (RAG) framework that initializes a diffusion-based planning policy by retrieving the most relevant expert demonstrations from the training dataset. By interpolating between current observations and retrieved examples through a denoising process, our approach enables fine-grained control and safe behavior across diverse scenarios, leveraging the strong prior provided by the retrieved scenario. Another key insight we produce is that a task-relevant retrieval model trained with planning-based objectives results in superior planning performance in our framework compared to a task-agnostic retriever. Experimental results demonstrate improved generalization to long-tail events and enhanced trajectory diversity compared to standard learning-based planners -- we observe a 40% reduction in collision rate on the Waymo Open Motion dataset with RAG.
title RealDrive: Retrieval-Augmented Driving with Diffusion Models
topic Robotics
Artificial Intelligence
url https://arxiv.org/abs/2505.24808