Control the 17-DOF ORCA Hand across hardware and MuJoCo simulation backends.
npx skills add https://github.com/broomva/skills --skill orcahand
Please help me install this Agent Skill. Skill: orcahand Source: https://github.com/broomva/skills/tree/main/skills/robotics/orcahand Command: npx skills add https://github.com/broomva/skills --skill orcahand
This skill provides a comprehensive interface for managing the ORCA Hand, a 17-DOF tendon-driven robotic hand, bridging the gap between physical hardware control, MuJoCo simulation, and reinforcement learning training.
Use the orcahand skill to bootstrap your workspace and detect the connected hardware.
High-intent search queries and answers about installing and using this skill.
This skill provides a full-stack control interface for the 17-DOF ORCA Hand, bridging physical hardware control, MuJoCo simulation, and reinforcement learning training using identical typed schemas for both backends.
To train grasp policies for a dexterous robotic hand, use the Evaluator-Governed Recursive Improvement loop within MuJoCo simulation, which maintains safety shields for range of motion, current, and temperature during policy optimization.
Yes, you can teleoperate a robotic hand using Apple Vision Pro or Rokoko gloves by integrating with orca_core, orca_sim, and orca_retargeter to maintain safety shields and multi-rate control loops.
Yes, setting up robotic hand hardware requires the Dynamixel SDK, PyTorch, mujoco, gymnasium, numpy, pytorch_kinematics, and pyyaml to assemble, calibrate, and control the physical hardware and train RL policies.
A unified plant interface handles safety for robotic hand reinforcement learning by implementing robust safety shields for range of motion, current, and temperature, combined with an Evaluator-Governed Recursive Improvement loop for safe policy optimization.
Limitations of using MuJoCo for robotic hand teleoperation include the strict requirement to integrate orca_core, orca_sim, and orca_retargeter to maintain safety shields and multi-rate control loops when switching to physical hardware.
jonnabioEvaluate NVIDIA Cosmos Policy in LIBERO and RoboCasa simulation environments.
Lord1EgyptTrain and evaluate reinforcement learning agents with Stable Baselines3 algorithms.
VaggelisKaBuild and distribute custom Expo development clients via EAS for iOS and Android.
robotlearning123Train and evaluate reinforcement learning agents with PyTorch and Gymnasium.
