Learning from Mistakes: Post-Training for Driving VLA with Takeover Data

TL;DR

This paper introduces TakeVLA, a novel post-training framework for vision-language-action in autonomous driving that improves safety margins and performance by proactive mistake learning and active scenario exploration.

Contribution

The paper presents two innovations: pre-takeover language supervision for proactive mistake learning and Scenario Dreaming for active exploration in reconstructed scenarios.

Findings

Achieves state-of-the-art performance on Bench2Drive benchmark.

Surpasses baseline by 4.93 in driving score.

Increases safety margin with 11.76% higher average TTC.

Abstract

Current Vision-Language-Action (VLA) paradigms in end-to-end autonomous driving rely on offline training from static datasets, leaving them vulnerable to distribution shift. Recent post-training methods use takeover data to mitigate this by augmenting the dataset with high-quality expert takeover samples, yet they suffer from two key limitations: supervision restricted to the period after the takeover moments leads to policies with limited safety margins, and passive preference optimization lacks active exploration for optimal performance. In this paper, we propose TakeVLA, a novel VLA post-training framework that overcomes these shortcomings through two complementary innovations. First, we introduce pre-takeover language supervision, which allows the VLA to learn from mistakes proactively. By explicitly teaching the model about what to do in error-prone situations, we cultivate a precautionary mindset that anticipates hazards early and substantially enlarges safety margins. Second, we propose Scenario Dreaming, a reinforcement fine-tuning paradigm that operates in reconstruceted takeover scenarios, encouraging active exploration beyond mere preference fitting. Experiments on the Bench2Drive benchmark demonstrate that TakeVLA achieves state-of-the-art closed-loop performance, surpassing the strong VLA baseline SimLingo by 4.93 in driving score, with an enhanced safety margin as evidenced by an 11.76% increase in average TTC.