PyRoki: A Modular Toolkit for Robot Kinematic Optimization

We unify problems like inverse kinematics, trajectory optimization, and motion retargeting using composable kinematic variables and costs. PyRoki aims to support a broad variety of robots and tasks, and runs on CPU, GPU, and TPU. Abstract -- Robot motion can have many goals. Depending on the task, we might optimize for pose error, speed, collision, or similarity to a human demonstration. Unlike existing tools, it is also cross-platform: optimization runs natively on CPU, GPU, and TPU. In this paper, we present (i) the design and implementation of PyRoki, (ii) motion retargeting and planning case studies that highlight the advantages of PyRoki's modularity, and (iii) optimization benchmarking, where PyRoki can be 1.4-1.7x I NTRODUCTION Numerical optimization is the standard solution for many tasks in robot kinematics. Using objectives like pose error [8], smoothness [9], and similarity to a human demonstration [6, 10] the robotics community has built diverse optimization software for tasks such as inverse kinematics (IK) [1, 3, 11-13], trajectory optimization [4, 5, 14-19], and Equal contribution.