CritiqueDriveVLM: From Verifier-Guided Reinforcement Learning to Latent Thought Distillation for Autonomous Driving

• URL: http://arxiv.org/abs/2607.04179v1
• Date: Sun, 05 Jul 2026 08:51:47 GMT
• Title: CritiqueDriveVLM: From Verifier-Guided Reinforcement Learning to Latent Thought Distillation for Autonomous Driving
• Abstract summary: Vision-Language Models (VLMs) show immense potential in autonomous driving.<n>CritiqueDriveVLM is a novel unified three-stage framework internalizing reasoning directly into the VLM.<n>By aligning the Student's latent representations with the Teacher's fully converged reasoning states, we compress deep logical capabilities into a fast, CoT-free System-1 Student.
• Score: 14.150596203774384
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: End-to-end Vision-Language Models (VLMs) show immense potential in autonomous driving. However, standard Supervised Fine-Tuning (SFT) often suffers from reasoning hallucinations and conservative biases. While traditional tool-augmented frameworks and Chain-of-Thought (CoT) approaches mitigate these issues, they incur exorbitant token consumption and unacceptable latency, rendering real-time deployment impractical. To resolve this reliability-efficiency trade-off, we propose CritiqueDriveVLM, a novel unified three-stage framework internalizing reasoning directly into the VLM. First, we introduce Critique-Driven Multi-Turn Reinforcement Learning (RL) guided by a multi-dimensional verifier. By providing granular scalar feedback and a multi-turn penalty, we force the policy to internalize logical deduction, cultivating a robust System-2 Teacher that achieves high accuracy without fragile external tools. Subsequently, we propose Latent Thought Distillation to overcome the latency bottleneck. By aligning the Student's latent representations with the Teacher's fully converged reasoning states, we compress deep logical capabilities into a fast, CoT-free System-1 Student. Extensive experiments on the widely-used DriveLMM-01 benchmark demonstrate remarkable improvements. Compared to the base model, our tool-free Teacher significantly boosts Multiple Choice Quality (MCQ) from 55.54% to a state-of-the-art 76.54%. Crucially, our distilled Student preserves competitive reasoning depth while drastically minimizing generation length to an average of merely 28 tokens. This slashes inference latency by 88% (from 3482 ms to 416 ms), paving a highly robust pathway for low-latency autonomous driving.Our source code is available at https://github.com/MICLAB-BUPT/CritiqueDriveVLM.

Related papers

• Teaching Thinking Models to Reason with Tools: A Full-Pipeline Recipe for Tool-Integrated Reasoning [59.74608632210439]
  We show how to inject natural tool-use behavior into a strong thinking model without sacrificing its no-tool reasoning ability.<n>Our recipe yields models that achieve state-of-the-art performance in a wide range of benchmarks among open-source models.
  arXiv  Detail & Related papers  (2026-05-07T14:23:21Z)
• Drive-KD: Multi-Teacher Distillation for VLMs in Autonomous Driving [26.97190983537793]
  We propose Drive-KD, a framework that decomposes autonomous driving into a "perception-reasoning-planning" triad.<n>We identify layer-specific attention as the distillation signal to construct capability-specific single-teacher models that outperform baselines.<n>Experiments show that our distilled InternVL3-1B model, with 42 times less GPU memory and 11.4 times higher throughput, achieves better overall performance than the pretrained 78B model.
  arXiv  Detail & Related papers  (2026-01-29T05:41:24Z)
• ThinkDrive: Chain-of-Thought Guided Progressive Reinforcement Learning Fine-Tuning for Autonomous Driving [14.981675960513606]
  Existing methods suffer from unstructured reasoning, poor generalization, and misalignment with human driving.<n>We propose ThinkDrive, a CoT guided progressive RL fine-tuning framework for autonomous driving that synergizes explicit reasoning with difficulty-aware adaptive policy optimization.<n>The results show that ThinkDrive outperforms strong RL baselines by 1.45%, 1.95%, and 1.01% on exam, easy-exam intensity and accuracy, respectively.
  arXiv  Detail & Related papers  (2026-01-08T08:30:36Z)
• Addressing Overthinking in Large Vision-Language Models via Gated Perception-Reasoning Optimization [56.59356959631999]
  Gated Perception-Reasoning Optimization (GPRO) is a meta-reasoning controller that dynamically routes computation among three decision paths.<n>GPRO substantially improves both accuracy and efficiency, outperforming recent slow-thinking methods.
  arXiv  Detail & Related papers  (2026-01-07T23:05:17Z)
• OmniDrive-R1: Reinforcement-driven Interleaved Multi-modal Chain-of-Thought for Trustworthy Vision-Language Autonomous Driving [11.793091014248025]
  We introduce OmniDrive-R1, an end-to-end VLM framework designed for autonomous driving.<n>Our core innovation is an Reinforcement-driven visual grounding capability.<n>This capability is enabled by our pure two-stage reinforcement learning training pipeline and Clip-GRPO algorithm.
  arXiv  Detail & Related papers  (2025-12-16T03:19:28Z)
• MindDrive: A Vision-Language-Action Model for Autonomous Driving via Online Reinforcement Learning [51.20229133553804]
  Current Vision-Language-Action (VLA) paradigms in autonomous driving primarily rely on Imitation Learning (IL)<n>Online Reinforcement Learning offers a promising pathway to address these issues through trial-and-error learning.<n>We propose MindDrive, a VLA framework comprising a large language model (LLM) with two distinct sets of LoRA parameters.<n>By feeding trajectory-level rewards back into the reasoning space, MindDrive enables trial-and-error learning over a finite set of discrete linguistic driving decisions.
  arXiv  Detail & Related papers  (2025-12-15T18:31:32Z)
• Directional Reasoning Injection for Fine-Tuning MLLMs [51.53222423215055]
  Multimodal large language models (MLLMs) are rapidly advancing, yet their reasoning ability often lags behind that of strong text-only counterparts.<n>Existing methods to bridge this gap rely on supervised fine-tuning over large-scale multimodal reasoning data or reinforcement learning.<n>We propose Directional Reasoning Injection for Fine-Tuning (DRIFT) to solve this problem.
  arXiv  Detail & Related papers  (2025-10-16T18:06:46Z)
• AdaThinkDrive: Adaptive Thinking via Reinforcement Learning for Autonomous Driving [21.10362636088305]
  Chain of Thought (CoT) has been widely adopted in Vision Language Action (VLA) models.<n>We propose AdaThinkDrive, a novel VLA framework with a dual mode reasoning mechanism inspired by fast and slow thinking.
  arXiv  Detail & Related papers  (2025-09-17T07:35:39Z)
• Consistent Paths Lead to Truth: Self-Rewarding Reinforcement Learning for LLM Reasoning [87.7836502955847]
  We propose a novel self-rewarding reinforcement learning framework to enhance Large Language Model (LLM) reasoning.<n>Our key insight is that correct responses often exhibit consistent trajectory patterns in terms of model likelihood.<n>We introduce CoVo, an intrinsic reward mechanism that integrates Consistency and Volatility via a robust vector-space aggregation strategy.
  arXiv  Detail & Related papers  (2025-06-10T12:40:39Z)
• Beyond 'Aha!': Toward Systematic Meta-Abilities Alignment in Large Reasoning Models [86.88657425848547]
  Large reasoning models (LRMs) already possess a latent capacity for long chain-of-thought reasoning.<n>We explicitly align models with three meta-abilities: deduction, induction, and abduction, using automatically generated, self-verifiable tasks.<n>Our three stage-pipeline individual alignment, parameter-space merging, and domain-specific reinforcement learning, boosts performance by over 10% relative to instruction-tuned baselines.
  arXiv  Detail & Related papers  (2025-05-15T17:58:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.