Haosong Peng, Xiangyang Liu, Zhengshen Zhang, Yunfeng Fan, Tianyu Qi, Xin Xia, Yufeng Zhan and Wenchao Xu (arXiv) · Unified TCP Tracker

TCP4D: A Unified Tool-Center-Point Tracking and Generation Model for Robot Manipulation

TCP4D is a unified model that uses a shared 4D scene representation to both track tool-center-point (TCP) motion from monocular video and generate future TCP trajectories for robot manipulation. Its tracker reports absolute trajectory errors of 7.60 mm on DROID and 16.07 mm on RoboLab, and the generator reaches 69.19% mean success on RoboTwin. The paper also introduces a cross-embodiment augmentation pipeline that retargets existing RGB videos to new robots, with joint training improving target-robot success by 41.5% over augmented-only training.

Open the source ↗

Similar Agents & Automation releases

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.