Back to AI News
arxiv

WACO Makes Few-Shot Robot Policies Robust to Camera and Base Shifts

This arXiv paper argues that camera bumps and robot base re-docking are the same failure: a memorised camera-to-base extrinsic change in vision-language-action policies. It introduces WACO, a world-anchored canonicalisation framework that uses the arm's forward kinematics as a metric calibration target and the static scene as a world anchor, estimating both shifts from a single RGB-D frame without ground-truth extrinsics or extra data. The authors also release LIBERO-PoseShift, described as the first benchmark for RGB VLAs with graded camera, base and compound shifts. WACO reportedly raises shifted success by 13.2 points in simulation and 36.7 points on a UR5e real robot, recovering most of the gap to a ground-truth oracle without perturbed training data or test-time calibration.

AI Brief

It offers a practical calibration-free fix and a new benchmark for a common, under-addressed failure mode in deployed robot manipulation policies.

arXiv:2611.24360v1 Announce Type: cross Abstract: Fine-tuning a vision-language-action (VLA) model on a few dozen demonstrations is the standard way to put a manipulation policy on a robot, yet the resulting policy silently assumes that the camera and the robot base stay exactly where they were during data collection: a bumped tripod or a base re-docked a few centimetres away makes it reach for the wrong place. We show that both failures are one phenomenon, a change of the camera-to-base extrinsic that the policy has memorised, and that the robot carries everything needed to undo it. We introduce WACO, a world-anchored canonicalization framework for VLA models: the arm, placed by forward kinematics at the measured joint angles, serves as a metric calibration target, and the static scene serves as the world anchor that tells a moved camera from a moved robot. From a single RGB-D frame, WACO estimates both shifts without ground-truth extrinsics or extra data, canonicalises the image to the calibration viewpoint and the state, action and start pose to the calibration base frame, and fine-tunes the unmodified policy under the artefacts of its own estimator. We also introduce LIBERO-PoseShift, the first benchmark for RGB VLAs with graded camera, base and compound shifts. Across LIBERO-PoseShift, the LIBERO-Plus camera track and four real-world tasks on a UR5e robot, WACO consistently outperforms the standard few-shot recipe and prior viewpoint-robust methods, raising shifted success by 13.2 points in simulation and by 36.7 points on the real robot, recovering 90% of the simulation gap to a ground-truth oracle and 87% of the real-robot gap, and establishing a new state of the art among methods that use no perturbed training data, with no test-time calibration. Project page: https://peacelwh.github.io/WACO/.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!