Event
EgoThumb: A Vision-Driven Open-Source Robotic Thumb
1 reports · 1 sources
Jianyu Hou, Jiawei Fu, Weikang Wan, and other authors released a paper proposing EgoThumb, a vision-driven additional robotic thumb, and claim it can autonomously complete grasping and holding assistance. The system uses a vision-based imitation learning policy that maps first-person images from a palm camera directly to continuous joint positions. On the hardware side, EgoThumb uses a compact dual-tendon-driven finger, moving the actuation unit to the forearm to reduce distal weight and achieve precise active joint control. The authors claim the design can generalize across different users and environments; both hardware and software have been open-sourced, and the paper has been accepted by IROS.
Report timeline
- EgoThumb: A Vision-Driven Supernumerary Robotic Thumb for Autonomous Grasping and Holding Assistance
EgoThumb is a fully open-source, vision-driven supernumerary robotic thumb that uses a visual imitation learning policy to map first-person images from a palm-facing camera directly to continuous joint positions, enabling autonomous grasping and holding assistance. Its compact dual tendon-driven finger relocates the actuation unit to the forearm to reduce distal weight and support precise active joint control. The authors state that the design generalizes across different users and environments, that both hardware and software are open-sourced, and that the paper has been accepted to IROS.
arXiv cs.RO ·