Differentiable Gaussian Process Motion Planning
This work addresses a practical issue in robotics motion planning by making parameter tuning data-driven, though it is incremental as it builds on existing GPMP methods.
The paper tackles the problem of manually setting parameters in Gaussian Process Motion Planning (GPMP) algorithms, which can significantly affect performance, by proposing a differentiable extension to GPMP2 that enables end-to-end learning from data to automatically adapt these parameters.
Modern trajectory optimization based approaches to motion planning are fast, easy to implement, and effective on a wide range of robotics tasks. However, trajectory optimization algorithms have parameters that are typically set in advance (and rarely discussed in detail). Setting these parameters properly can have a significant impact on the practical performance of the algorithm, sometimes making the difference between finding a feasible plan or failing at the task entirely. We propose a method for leveraging past experience to learn how to automatically adapt the parameters of Gaussian Process Motion Planning (GPMP) algorithms. Specifically, we propose a differentiable extension to the GPMP2 algorithm, so that it can be trained end-to-end from data. We perform several experiments that validate our algorithm and illustrate the benefits of our proposed learning-based approach to motion planning.