Reflect-VLA: Internalizing Reflective Guidance via Dual-Path On-Policy Self-Distillation for Autonomous Driving
Reinforcement Learning (RL) substantially improves planning performance in autonomous-driving Vision-Language-Action (VLA) models.
ProofPaper ↗
Key points
- We observe that Reflective Guidance turns trajectory evaluation into explicit diagnoses and corrective instructions, substantially improving planning accuracy.
- We propose Reflect-VLA, which treats Reflective Guidance as training-time privileged information and internalizes its planning benefits through Dual-Path On-Policy Self-Distillation (Dual-OPSD).
- At inference time, Reflect-VLA generates trajectories in a single pass without explicit guidance or iterative refinement.
- Experiments on NAVSIM v1/v2 and Bench2Drive demonstrate competitive planning performance while reducing training time by over 90% compared with RL.
Sources (1)
- [1]Reflect-VLA: Internalizing Reflective Guidance via Dual-Path On-Policy Self-Distillation for Autonomous DrivingarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 11:52 AM
Reinforcement Learning (RL) substantially improves planning performance in autonomous-driving Vision-Language-Action (VLA) models.
We observe that Reflective Guidance turns trajectory evaluation into explicit diagnoses and corrective instructions, substantially improving planning accuracy.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 9, 2026Dual-Stream Memory for Vision-Language-Action Policies
- Oct 9, 2026Mitigating Capability Degradation in LLM Self-Improvement via Representation Geometry Regularization
- Oct 8, 2026Rewiring Semantics, Dynamics, and Control: A Simple yet Effective Action-Centric Tri-Stream Transformer
- Oct 8, 2026Distilling Routed 3D Privilege for Spatial Reasoning in Vision-Language Models
- Oct 8, 2026Embodied Turing Machines: Stateful Code for Robot Recursive Self-Improvement
- Oct 7, 2026Explicit Geometric Chain-of-Thought for Vision-Language-Action in Autonomous Driving