AdaDrive:面向语言引导自动驾驶的自适应慢-快系统
AdaDrive: Self-Adaptive Slow-Fast System for Language-Grounded Autonomous Driving
AI总结:
AdaDrive提出自适应慢-快框架,通过动态激活与连续融合策略决定LLM在自动驾驶决策中的参与时机和程度,在保证实时性的同时提升驾驶精度。
摘要:
将大语言模型(LLMs)有效集成到自动驾驶中,需要在利用高层推理与保持实时效率之间取得平衡。现有方法要么过于频繁地激活LLMs,导致过高的计算开销,要么采用固定调度,无法适应动态驾驶条件。为应对这些挑战,我们提出AdaDrive,一个自适应协作的慢-快框架,能够最优地决定LLMs何时以及如何参与决策。(1)何时激活LLM:AdaDrive采用一种新颖的自适应激活损失,基于比较学习机制动态决定LLM的调用,确保仅在复杂或关键场景下激活。(2)如何集成LLM辅助:AdaDrive没有采用刚性的二元激活方式,而是引入一种自适应融合策略,根据场景复杂度和预测置信度调节连续的、可缩放的LLM影响,确保与传统规划器无缝协作。通过这些策略,AdaDrive提供了一个灵活的、上下文感知的框架,在不牺牲实时性能的前提下最大化决策精度。在语言引导自动驾驶基准上的大量实验表明,AdaDrive在驾驶精度和计算效率方面均达到了最先进的性能。代码可在https://github.com/ReaFly/AdaDrive获取。
英文摘要:
Effectively integrating Large Language Models (LLMs) into autonomous driving requires a balance between leveraging high-level reasoning and maintaining real-time efficiency. Existing approaches either activate LLMs too frequently, causing excessive computational overhead, or use fixed schedules, failing to adapt to dynamic driving conditions. To address these challenges, we propose AdaDrive, an adaptively collaborative slow-fast framework that optimally determines when and how LLMs contribute to decision-making. (1) When to activate the LLM: AdaDrive employs a novel adaptive activation loss that dynamically determines LLM invocation based on a comparative learning mechanism, ensuring activation only in complex or critical scenarios. (2) How to integrate LLM assistance: Instead of rigid binary activation, AdaDrive introduces an adaptive fusion strategy that modulates a continuous, scaled LLM influence based on scene complexity and prediction confidence, ensuring seamless collaboration with conventional planners. Through these strategies, AdaDrive provides a flexible, context-aware framework that maximizes decision accuracy without compromising real-time performance. Extensive experiments on language-grounded autonomous driving benchmarks demonstrate that AdaDrive state-of-the-art performance in terms of both driving accuracy and computational efficiency. Code is available at https://github.com/ReaFly/AdaDrive.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。