DiffSim2Real: Deploying Quadrupedal Locomotion Policies Purely Trained in Differentiable Simulation
This work addresses the sim-to-real transfer challenge for robotics, enabling more efficient training without physical accuracy trade-offs, though it is incremental in applying differentiable simulation to a specific domain.
The paper tackled the problem of transferring locomotion policies trained in differentiable simulators to real quadrupedal robots, achieving successful real-world deployment with policies trained exclusively using analytic gradients from a smooth contact model.
Differentiable simulators provide analytic gradients, enabling more sample-efficient learning algorithms and paving the way for data intensive learning tasks such as learning from images. In this work, we demonstrate that locomotion policies trained with analytic gradients from a differentiable simulator can be successfully transferred to the real world. Typically, simulators that offer informative gradients lack the physical accuracy needed for sim-to-real transfer, and vice-versa. A key factor in our success is a smooth contact model that combines informative gradients with physical accuracy, ensuring effective transfer of learned behaviors. To the best of our knowledge, this is the first time a real quadrupedal robot is able to locomote after training exclusively in a differentiable simulation.