Closing the Real-to-Sim-to-Real Loop for Physical AI

Building Physical AI systems requires more than simulating idealized environments—it requires a way to connect real robot behavior, simulation, data generation, and deployment. In this livestream, Industrial Next introduces the R2S2R Arena, an open coalition for advancing real-to-sim-to-real research with academia and the broader robotics community.
Industrial Next will outline the challenge and the Arena’s benchmark tasks at sim2world.ai. Versor will share how robot trajectories can drive reconstruction, diverse data generation, and reinforcement learning to extend what can be simulated. AllSides will show how physically grounded, relightable USD assets and scalable variant generation can support training and testing. Together, the speakers will show how these pieces contribute to closing the Sim2Real gap.

You’ll Learn:

• Why real-to-sim-to-real workflows are important for Physical AI development
• How robot trajectories can support reconstruction, data generation, and reinforcement learning
• How scalable USD assets and benchmark tasks can support training and evaluation
• Why open collaboration across industry and academia matters for advancing robotics research