PRIMP: PRobabilistically-Informed Motion Primitives for Efficient Affordance Learning from Demonstration

Sipu Ruan, Weixiao Liu, Xiaoli Wang, Xin Meng, Gregory S Chirikjian

Keywords: Learning from Demonstration, Motion Planning, Service Robots

TL;DR: This paper proposes a robotic system that connects learning-from-demonstration, motion planning and robot imagination via physics simulation.

Abstract:

This paper proposes a learning-from-demonstration method using probability densities on the workspaces of robot manipulators. The method, named “PRobabilistically-Informed Motion Primitives (PRIMP)”, learns the probability distribution of the end effector trajectories in the 6D workspace that includes both positions and orientations. It is able to adapt to new situations such as novel via-point poses with uncertainty and a change of viewing frame. The method itself is robot-agnostic, in which the learned distribution can be transferred to another robot with the adaptation to its workspace density. The learned trajectory distribution is then used to guide an optimization-based
motion planning algorithm to further help the robot avoid novel obstacles that are unseen during the demonstration process. The proposed methods are evaluated by several sets of benchmark experiments. PRIMP runs more than 5 times faster than the compared existing probabilistic methods while generalizing trajectories more than twice as close to both the demonstrations and novel desired poses. It is then combined with our robot imagination method that learns object affordances, illustrating the applicability of PRIMP to learn tool use through physical experiments.