Robot policies improve action choices using past steps for safer moves
Looking Back to Move Forward: Temporal Verification for Generative Robot Policies
Robotics
Summary
Robots that learn from many example actions sometimes make mistakes that build up over time, leading them into unfamiliar situations they can’t easily fix. The researchers developed a method called Temporal Verification that helps the robot check candidate future actions while considering what it did just before, not just the current moment. This way, the robot can pick smoother, more reliable action sequences and avoid getting stuck. Their tests showed that this approach improves success and makes robot behavior more natural.
What this means in practice
- •For robotics engineers: Enhance autonomous robot controllers by verifying candidate actions using past trajectories for smoother and more reliable task execution.
- •For automation system designers: Improve industrial robot manipulators’ safety and efficiency by integrating temporally aware action verification during operation.
Authors
Haoxuan Wang, Wayne Wu, Yan Yan, Bolei Zhou
Abstract
Generative policies have emerged as a promising paradigm for robot learning, combining expressive generative action modeling with scalable imitation learning from large demonstration corpora. However, heterogeneous demonstrations can induce suboptimal action chunks whose errors compound over time, eventually driving the robot into out-of-distribution states from which recovery is difficult. Action verification offers a test-time scaling strategy for mitigating this failure mode by sampling multiple candidate actions and using a verifier to select one for execution. Existing approaches, however, remain temporally myopic and costly to train, evaluating candidates from the current observation alone without accounting for trajectory continuity and often relying on large verifiers and additional expert demonstrations. In this paper, we introduce Temporal Verification (TeV), an efficient temporally aware action verification framework for flow-matching VLAs. TeV first learns a temporal token that summarizes recent observation--action history, enabling candidate chunks to be evaluated as continuations of the execution trajectory rather than as isolated predictions. Conditioned on this token, TeV constructs positive--negative pairs without additional expert demonstrations or preference annotations and trains an energy-based verifier contrastively to assign lower energy to higher-quality, trajectory-consistent action chunks. Beyond post-hoc ranking, TeV further uses the learned energy landscape to guide intermediate flow samples toward lower-energy regions, improving candidates before final selection. Extensive experiments in simulation and real-world settings demonstrate that TeV provides reliably ranks action candidates, improves task success rates, and produces smoother execution trajectories.