Imitation Learning with Baxter Robot using Hi-Fives

Abstract

The goal of this paper is to successfully learn hi-fives for human-robot interaction. The proposed research work has used the Imitation Learning approach by incorporating Bayesian Interaction Primitives [1]. Through expert-guided demonstrations, the robot has been trained to learn relationships between human and robot trajectories. The research study has demonstrated that, the robot is able to complete the interaction with a human and successfully issue a hi-five. Also, the Bayesian Interaction Primitives are implemented to teach a Baxter Robot to give hi-five through imitation learning. Additionally, the trajectories are compared with human biomechanics data.