Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces VERITAS, a generator-verifier framework that enhances generalist robot policies through inference-time steering and self-improvement. By utilizing a pre-trained policy as a generator and a gradient-free visual verifier for action evaluation, the framework enables significant performance improvements without the need for additional training data. The findings reveal that verified rollouts not only outperform traditional generalist approaches but also allow for effective offline policy improvement, achieving efficiency comparable to expert demonstrations without human intervention.
Inference-time verification can boost robotic policy performance without extra training, rivaling expert demonstrations in efficiency.
Robots deployed in the real world should learn from their experience and improve over time. This requires a mechanism of practicing and learning from feedback. In this paper, we propose VERITAS, a generator-verifier framework for generalist robot policies for inference-time policy steering and self-improvement. We use a pre-trained generalist robot policy as a ``generator'' and pair it with a gradient-free ``visual verifier'' that evaluates actions at inference time. This framework enables inference-time steering that improves policy performance without additional training. We demonstrate that inference-time verification consistently outperforms vanilla generalists without training on additional demonstration data. Additionally, we demonstrate that the verified rollouts provide effective supervision for offline policy improvement: policies fine-tuned on verified self-generated trajectories achieve consistent performance gains. Notably, we find that post-training with verified rollouts achieves comparable efficiency to expert demonstrations, while requiring no human interventions. Our results highlight inference-time verification as a practical and scalable mechanism for improving robotic policies during deployment.