Robosuite

Robosuite is a modular simulation framework for robot learning developed by the ARISE Initiative at Stanford University. Built on MuJoCo physics, robosuite provides a standardized suite of manipulation tasks with a clean, modular API that enables researchers to compose robots, controllers, camera configurations, and task objects. The framework supports 8 robot models (Franka Panda, Sawyer, UR5e, IIWA, Jaco, Kinova3, Fetch, and humanoids) and includes 50+ manipulation tasks ranging from simple reaching and lifting to complex assembly and nut-and-screw insertion. Each task includes multiple difficulty levels and domain-randomization parameters. Robosuite integrates tightly with robomimic for imitation learning workflows, providing demonstrations, reward functions, and standardized evaluation protocols. It is widely used in both academia and industry as a benchmark for manipulation learning research.