Project SuperDex Tackles Soft-Object Simulation

SuperDex targets one of robotics simulation’s nastiest gaps: stable, useful contact with deformable objects. Its unified stack is promising, but the early-stage learning tools and limited independent validation mean developers should benchmark it against established simulators before committing.

• –Implicit simulation and volumetric soft actors aim to improve stability during manipulation and collisions
• –Robotics, Studio, and Lab modules cover asset creation, control, and policy training in one workflow
• –Apache-2.0 licensing makes the core platform accessible for research and commercial experimentation
• –Meta’s zero-shot shape-sorting demonstration suggests meaningful sim-to-real potential
• –SuperDex Lab remains early preview, while the teleoperation stack is planned for Q4 2026