Adaptive Gaussian Process based Stochastic Trajectory Optimization for Motion Planning

TL;DR

This paper introduces an advanced motion planning algorithm called iAGP-STO that combines Gaussian processes, adaptive stochastic optimization, and incremental planning to improve efficiency, safety, and reliability in complex robotic environments.

Contribution

It develops a novel integrated framework combining L-reestimation, stochastic trajectory learning, and incremental planning for improved robot motion planning.

Findings

Outperforms existing methods in efficiency and reliability benchmarks.

Demonstrates successful application on various robotic platforms.

Shows significant improvements in safety and computation time.

Abstract

We propose a new formulation of optimal motion planning (OMP) algorithm for robots operating in a hazardous environment, called adaptive Gaussian-process based stochastic trajectory optimization (AGP-STO). It first restarts the accelerated gradient descent with the reestimated Lipschitz constant (L-reAGD) to improve the computation efficiency, only requiring 1st-order momentum. However, it still cannot infer a global optimum of the nonconvex problem, informed by the prior information of Gaussian-process (GP) and obstacles. So it then integrates the adaptive stochastic trajectory optimization (ASTO) in the L-reestimation process to learn the GP-prior rewarded by the important samples via accelerated moving averaging (AMA). Moreover, we introduce the incremental optimal motion planning (iOMP) to upgrade AGP-STO to iAGP-STO. It interpolates the trajectory incrementally among the previously optimized waypoints to ensure time-continuous safety. Finally, we benchmark iAGP-STO against the numerical (CHOMP, TrajOpt, GPMP) and sampling (STOMP, RRT-Connect) methods and conduct the tuning experiment of key parameters to show how the integration of L-reAGD, ASTO, and iOMP elevates computation efficiency and reliability. Moreover, the implementation of iAGP- STO on LBR-iiwa, multi-AGV, and rethink-Baxter demonstrates its application in manipulation, collaboration, and assistance.