Search papers, labs, and topics across Lattice.
JitterVLA is a novel benchmark designed to isolate and evaluate the impact of robot controllers on motion quality in vision-language-action (VLA) driven manipulation, addressing the critical issue of abrupt motion caused by delayed and irregular command arrival. By replaying identical command and arrival traces through alternative controllers, JitterVLA quantifies motion smoothness, demonstrating that interpolation significantly reduces modeled p99 jerk across various VLA pipelines and human motion data, with minimal tracking cost. This benchmark enables robust pre-screening of controllers to ensure safer human-robot interaction before deployment in real-world scenarios.
Robot controllers can drastically alter motion smoothness, with simple interpolation reducing peak jerk by 83% in VLA-driven manipulation without significant tracking degradation.
Vision-language-action (VLA) policies issue sparse targets whose arrival at the robot controller is delayed and irregular, so a reasonable predicted action can still produce abrupt motion near people. Remedies are evaluated on different streams, delays and metrics. JitterVLA replays identical commands and arrival traces through alternative controllers, isolating the controller's effect. We apply it to synthetic reaches, hand trajectories from 20 people, three open VLA pipelines (838 episodes) and a physical UR3. Interpolation lowered modeled p99 jerk for every source, by 83% (615 to 106 m/s^3) on human motion with 200 ms delay, at a 2.0 cm tracking cost, and the effect replicated on 598 new episodes; jerk-limited generation lowered jerk further at a larger cost. Entropy-discontinuity associations had opposite, replicated signs in two pipelines, and entropy gating helped in one but not the other. On the UR3, interpolation lowered accelerometer-derived jerk by 3.0-14.0% in all six exploratory pairs, while a position-derived metric did not resolve the contrast. JitterVLA supports screening before human-centred studies; code and data will be released.
