The Schedule

RL training and benchmarks:

- Review of the training environments

- Standard training environments: Mujoco Ant, Humanoid - implementation, training result and live demo

- Advanced use cases: Shadow and Allegro Hand environments, OpenAI

- Trifinger robotic manipulation project: environment and benchmarks, how-to tips - implementation, sim-to-real results

Isaac Gym in academic labs:

- Robotic manipulation projects from the Intelligent Autonomous

Manipulation lab at Carnegie Mellon

- Sim-to-real transfer for legged robots locomotion on non-even terrain, ETH Zurich

Additional topics:

- Scalable Adaptive Domain Randomization with BayesSimIG

- GPU-accelerated operational space controllers

Adding controllers to GPU-pipeline

OSC controller -- what, why

OSC showcase -- Kendama

- Deformable object simulation via the finite element method

BioTac simulations, real-world validation

3D deformable object grasping

New frontiers:

- Roadmap for future development of Isaac Gym

- New PhysX simulation features: FEM soft-bodies and cloth, liquids, MPM simulation

- Training with vision; multi-camera path-traced sensors

Wrap up:

- Further resources, forums, code sharing sites

Breakout room for additional Q&A:

- Beginners’ room, RL room, physics room, vision room