Virdyn pitches mHand Studio for dexterous hand teleoperation

Virdyn says its mHand Studio software is designed to turn human hand motion into robot-ready data for dexterous teleoperation, simulation and training. The pitch targets robotics teams that need safer testing, cleaner datasets and a faster path from human demonstrations to embodied AI systems.

Why it matters: - Dexterous manipulation remains one of the hardest problems in robotics because human hands use complex, task-specific motions that are difficult to capture and reuse. - Robotics teams need data that is not just recorded, but synchronized, retargeted and structured for training and hardware control. - Virdyn is positioning mHand Studio as a way to reduce costly trial-and-error on physical robots by moving more of the workflow into software and simulation.

What happened: - Virdyn introduced mHand Studio as a unified workflow for motion capture, dexterous-hand teleoperation, simulation, hardware control and motion-data collection. - The company says the system is built around the Virdyn mHand Pro Motion Capture Gloves. - Virdyn frames the software as a tool for robotics R&D teams working on humanoid manipulation, embodied AI and human-in-the-loop control. - The company published a product page for the gloves with more information here.

The details: - The mHand Pro gloves use a multi-sensor inertial architecture. - The gloves have 16 sensors that capture hand and finger movement. - The system outputs raw data, quaternion, Euler and BVH formats. - Virdyn says the platform supports multiple SDK integration options. - The software pipeline runs from motion capture to retargeting, 3D simulation, hardware teleoperation, data recording and verification. - The system stores teleoperation data in CSV files, including joint angles and timestamps. - The workflow is designed to separate retargeted motion data from hardware feedback data. - Virdyn says the software can help with robot motion analysis, debugging, demonstration replay and imitation learning workflows. - The company says the system integrates motion capture, visualization, hardware control and recording into one workflow. - The URDF model view lets engineers inspect linkages, joints and physical relationships before sending commands to hardware. - The software is intended to support tasks such as grasping, rotation, insertion, manipulation and assembly.

Between the lines: - The core challenge is not only collecting human hand motion, but converting that motion into a robot-compatible representation that accounts for different joint counts, link dimensions, degrees of freedom and mechanical limits. - Timing is a major issue in teleoperation, because capture, retargeting, visualization, hardware commands and recording can drift if they are not synchronized. - Simulation acts as a verification step before physical execution, which can help avoid collisions, joint-limit violations and other hardware damage. - Google DeepMind has also explored motion-capture data as a source for learning controllable motor actions, underscoring why structured demonstration data matters in this area. - Virdyn is arguing that each confirmed demonstration can become a reusable dataset asset, not just a one-off robot test.

What's next: - Virdyn is encouraging robotics teams to test dexterous tasks in simulation before running them on hardware. - The company is inviting teams with repeatable hand-manipulation workflows to contact Virdyn about mHand Studio and its teleoperation pipeline. - The broader bet is that cleaner human demonstration data will speed up robot learning and improve dexterous hand performance over time.

The bottom line: - Virdyn is betting that the next leap in humanoid and dexterous robotics will come from better human-motion data pipelines, not just better robot hands.