Low-cost upper-limb exoskeleton for teleoperating robot arms with real-time haptic torque feedback, from Ant Group-backed research.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Universal Manipulation Exoskeleton (UME) is a wearable upper-limb exoskeleton for robot teleoperation and data collection. It records whole-arm joint configurations and joint torque signals while giving the operator real-time haptic torque feedback. The goal is to collect demonstrations that include force and torque data, which many existing data collection pipelines lack, so that compliant robot policies can be learned.
According to the project page, UME is low-cost (the full system is listed at $1,900), lightweight and portable. An embedded IMU supports mobile manipulation teleoperation. A universal retargeting algorithm lets it drive different arms, including the 7DoF OpenArm, 7DoF Franka and 6DoF X-ARM. Operators can perform tasks such as unsheathing a sword while blindfolded using torque feedback alone.
The authors trained autonomous bimanual, whole-body, compliant policies purely on UME data, using an in-house dual-arm mobile manipulator (about $9,533). Tasks include fridge drink retrieval (157 demonstrations), space-constrained GPU picking (42), force-mediated box flipping (40) and visually occluded box pushing (26). The work is a research project from Stanford-affiliated authors, supported by Ant Group. The page lists no public code, hardware files or purchase option; a creator mentioned code is coming soon.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Wearable upper-body exoskeleton for teleoperating and teaching robots, recording arm motion and force/torque feedback. Robots learn household tasks from demonstrations; code is coming soon.
Closest neighbours by meaning, not just by category.
Research
Research framework that retargets human motion to humanoid robots using learned point cloud correspondences.
Model
Action-conditioned robotics world model fine-tuned from Cosmos-Predict2.5-2B using 2D skeleton conditioning across embodiments.
Research · UC San Diego
Research system using a teleoperated Unitree G1 humanoid for laparoscopic surgery, tested in dry-lab and live porcine studies.
Model · Xiaomi
Open-weights Vision-Language-Action robot foundation model from Xiaomi, pre-trained on 100K+ hours of UMI trajectories.
Research · BitRobot Foundation
Open dataset of 500+ hours of whole-body teleoperated Unitree G1 humanoid data recorded in real homes.
Model
Research framework that conditions one video world model on force, camera, hand pose and robot pose actions.
