# Update-Free On-Policy Steering via Verifiers

arXiv: https://arxiv.org/abs/2603.10282 · HTML: https://mlsanity.com/paper/2603.10282/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 10 Mar 2026.

## Built on by (most cited first, 5 of 5)

- [QPILOTS: Efficient Test-Time Q-Steering for Flow Policies](https://mlsanity.com/paper/2606.14801/) (2026)
- [Safe Embodied AI for Long-horizon Tasks: A Cross-layer Analysis of Robotic Manipulation](https://mlsanity.com/paper/2606.05660/) (2026)
- [Learning Process Rewards via Success Visitation Matching for Efficient RL](https://mlsanity.com/paper/2606.23640/) (2026)
- [Behavior Uncloning: Distilling Mode Redirection into Policy Weights without Inference-Time Steering](https://mlsanity.com/paper/2606.29201/) (2026)
- [DiVeR: Decision-Critical Verifier Learning for VLA Test-Time Scaling](https://mlsanity.com/paper/2610.04933/) (2026)

## Abstract (arXiv)

In recent years, Behavior Cloning (BC) has become one of the most prevalent methods for learning manipulation from human demonstrations. Despite their successes, BC policies are often brittle and struggle with precise manipulation. To overcome these issues, we propose UF-OPS, an Update-Free On-Policy Steering method that enables the robot to predict the success likelihood of its actions and adapt its strategy at execution time. We accomplish this by training verifier functions using policy rollout data obtained during an initial evaluation of the policy. These verifiers are subsequently used to steer the base policy toward actions with a higher likelihood of success. Our method improves the performance of black-box diffusion policies, without changing the base parameters, making it lightweight and flexible. We present results from both simulation and real-world data and achieve an average 49% improvement in success rate over the base policy across 5 real tasks.