Differentiable and Learnable Robot Models
This work provides a tool for robotics researchers to combine simulation and machine learning, but it is incremental as it builds on existing differentiable simulation efforts.
The authors tackled the problem of integrating data-driven methods with analytical rigid body computations by developing a library called Differentiable Robot Models, which implements differentiable and learnable kinematics and dynamics models in PyTorch, making it publicly available on GitHub.
Building differentiable simulations of physical processes has recently received an increasing amount of attention. Specifically, some efforts develop differentiable robotic physics engines motivated by the computational benefits of merging rigid body simulations with modern differentiable machine learning libraries. Here, we present a library that focuses on the ability to combine data driven methods with analytical rigid body computations. More concretely, our library \emph{Differentiable Robot Models} implements both \emph{differentiable} and \emph{learnable} models of the kinematics and dynamics of robots in Pytorch. The source-code is available at \url{https://github.com/facebookresearch/differentiable-robot-model}