Shypd.aiThe largest open index of AI toolsMuJoCo_RL_UR5 is an open-source MuJoCo/Gym environment for robot control using Reinforcement Learning. It enables pixel-wise prediction of grasp success chances for robotic pick and place tasks with visual observations.
Trending
About
MuJoCo_RL_UR5 provides a MuJoCo/Gym environment specifically designed for robot control through Reinforcement Learning. Its core task involves agents performing pixel-wise prediction of grasp success chances, utilizing purely visual observations. The environment supports training agents for pick and place tasks with a Universal Robots UR5 and a Robotiq S Model 3 Finger Gripper. Key features include a very large and multi-discrete action space, high sample-cost, and a binary reward system. It offers classes like MJ_Controller for basic robot control and GraspEnv for training RL agents, providing RGB and depth images as observations. The repository is no longer under active development but serves as a valuable resource for robotics researchers and reinforcement learning engineers.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
