Search papers, labs, and topics across Lattice.
This paper introduces CheckVLA, a method for verifying execution in long-horizon mobile manipulation tasks by utilizing a frozen action-conditioned world model to assess deviations from expected outcomes. The approach addresses the limitations of open-loop action execution, where errors can propagate without corrective feedback, by implementing a calibrated risk threshold to determine intervention points and employing latency-aware strategies for action replacement. On the RoboCasa365 benchmark, CheckVLA significantly outperformed traditional periodic replanning, achieving a 36.1% success rate and improving timely recall to 77.9% compared to lower performance from observation-only methods.
Action-conditioned verification can boost success rates in mobile manipulation tasks by over 8% while enhancing timely recall by nearly 30%.
Vision-language-action (VLA) policies commonly execute long-horizon mobile manipulation through open-loop action chunks, issuing multiple actions without receiving new high-level visual input. A committed chunk therefore implies how observations should evolve, but accidental deviations can violate this expectation while the remaining actions continue to propagate the error: commit-time policy confidence cannot react to a deviation that occurs after dispatch, and observation-only anomaly scores lack an action-conditioned reference for separating expected effects from unexplained changes. We propose CheckVLA, which verifies execution with a separately trained, frozen action-conditioned world model. A conformally calibrated risk threshold bounds the episode-level probability of an unnecessary first intervention and determines when to intervene, its exceedance controls how strongly the rewritten suffix retains the superseded chunk, latency-aware hard prefixing restricts replacement to actions that remain deployable, and an event-driven keyframe bank preserves evidence of prior progress across repairs. On RoboCasa365, under a common training recipe and a matched invocation budget, CheckVLA attains a 36.1% average success rate against 27.6% for periodic replanning (+8.5 points). At a matched 5% episode-level false-alarm target, action conditioning raises timely recall to 77.9%, against 48.6% for an observation-only control and 37.9% for an action-shuffled control. These simulation results support action-conditioned verification as a way to restore feedback during chunked execution while keeping the repair consistent with inference latency.
