Generate validated and visualizable ORCA hand MuJoCo RL environments from natural language.
Generate validated, visualizable ORCA v1 right-hand MuJoCo/Gymnasium RL environments from Chinese or English task descriptions. Use for hand gestures, picking up a box/cylinder/sphere, or placing one of those objects in a target region; reject tool use, insertion, stacking, external meshes, and two-hand requests.
