OSMO: a tactile glove for humans and robots
What open-source hardware makes possible, and the questions that remain about preserving human skill.
INTERACT ROBOTICS · Research Notes · Reviewed September 7, 2026
Can robots learn directly from human hands?
OSMO is an open-source glove with twelve three-axis magnetic tactile sensors across the fingers and palm. Published hardware, firmware, and assembly instructions support an approach in which humans and robots share similar tactile observations. This article covers the glove research, separate from the robotics workflow service of the same name.
What the research demonstrated.
The paper reports 72% success on sustained-contact wiping, trained only on human demonstrations with no robot demonstrations. It does not establish immediate transfer of all manual skills. It suggests that observation differences can be reduced for a particular sensor configuration and task.
References
- OSMO: Open-Source Tactile Glove for Human-to-Robot Skill Transfer ↗
Jessica Yin et al. / December 9, 2025. The 72% result applies to the wiping task evaluated in the paper.
- OSMO / Hardware, firmware, and data repository ↗
Use the official repository for build instructions and guidance on the released data.
Good sensors don’t guarantee good demonstrations.
A thicker fingertip can change how a small part is grasped. Different friction can change the force a person applies. Alongside sensor count and range, evaluate finger mobility, surface perception, and changes in motion during extended wear.
These are criteria we intend to validate in glove development, not evidence that our glove outperforms OSMO. We consider comparisons using the same person and task, bare-handed and gloved, covering success, time, force use, and subjective feel.
An interface that lets human skill come through.
Our goal goes beyond collecting more tactile readings. People should feel stiffness and slip, use their own skill, and leave a record of contact and response. We also need shared representations across different human and robot bodies. OSMO makes these questions approachable through reproducible hardware.