Learning What to Say to Your VLA: Mostly Harmless Vision Language Action Model Steering

arXiv:2606.12299 · cs.RO, cs.LG · Submitted 2026-06-10 · Read on arXiv

cs.RO, cs.LG

Submitted: 2026-06-10

Updated: 2026-10-16

Comments: CoRL 2026

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Steering — 3× in this paper · explained in 2 episodes
• Fine-Tuning — 1× in this paper · explained in 5 episodes
• Vision-Language-Action (VLA) Models — 1× in this paper · explained in 5 episodes

Sources

• VLAs are Confined yet Capable of Generalizing to Novel Instructions
• OpenVLA: An Open-Source Vision-Language-Action Model
• $\pi_{0.5}$: a Vision-Language-Action Model with Open-World Generalization
• ${\pi}_{0.7}$: a Steerable Generalist Robotic Foundation Model with Emergent Capabilities
• CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models
• MolmoAct: Action Reasoning Models that can Reason in Space
• MolmoAct2: Action Reasoning Models for Real-world Deployment
• Do What You Say: Steering Vision-Language-Action Models via Runtime Reasoning-Action Alignment Verification
• A Smooth Sea Never Made a Skilled SAILOR: Robust Imitation via Learning to Search
• Self-Improving Vision-Language-Action Models with Data Generation via Residual RL
• Steerable Vision-Language-Action Policies for Embodied Reasoning and Hierarchical Control
• Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignment
• Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training
• ReSteer: Quantifying and Refining the Steerability of Multitask Robot Policies
• SafeVLA: Towards Safety Alignment of Vision-Language-Action Model via Constrained Learning
• Mechanistic Finetuning of Vision-Language-Action Models via Few-Shot Demonstrations
• Mechanistic interpretability for steering vision-language-action models
• Observing and Controlling Features in Vision-Language-Action Models
• SAFE: Multitask Failure Detection for Vision-Language-Action Models
• DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset

Related papers

• FMT x: An Efficient and Asymptotically Optimal Extension of the Fast Marching Tree for Dynamic Replanning
• MPCFormer: A physics-informed data-driven approach for explainable socially-aware autonomous driving
• RoboLab: A High-Fidelity Simulation Benchmark for Analysis of Task Generalist Policies
• ARC: A Reasoning Recipe for Robot Foundation Models
• HRDexDB: A 4D Dexterous Grasping Dataset Across Human and Multiple Robot Embodiments
• APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies