RoboIRS: Inference-Time Internal Representation Steering for Generalist Robot Policies
cs.RO
Submitted: 2026-10-03
Updated: 2026-10-08
Project page: https://rollingoat.github.io/roboirs
Terminology
Sources
- LIBERO-PRO: Towards Robust and Fair Evaluation of Vision-Language-Action Models Beyond Memorization
- Steering Language Models With Activation Engineering
- Representation Engineering: A Top-Down Approach to AI Transparency
- Not All Features Are Created Equal: A Mechanistic Study of Vision-Language-Action Models
- Sparse Autoencoders Reveal Interpretable and Steerable Features in VLA Models
- Mechanistic interpretability for steering vision-language-action models
- Observing and Controlling Features in Vision-Language-Action Models
- Steering Robustness into World Action Models via Mechanistic Interpretability and Optimal Control
- $\pi_{0.5}$: a Vision-Language-Action Model with Open-World Generalization
- Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning
- The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
- Linear Representations of Sentiment in Large Language Models
- In-Context Learning Creates Task Vectors
- Function Vectors in Large Language Models
- Refusal in Language Models Is Mediated by a Single Direction
- Mechanistic Finetuning of Vision-Language-Action Models via Few-Shot Demonstrations
- Contrastive Conceptor Activation Steering (COAST): Unlocking Vision-Language-Action Models through Hidden States
- RoVer: Robot Reward Model as Test-Time Verifier for Vision-Language-Action Model
- Guided Action Flow: Value-Guided Sampling for Frozen Vision-Language-Action Policies
- PPGuide: Steering Diffusion Policies with Performance Predictive Guidance
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