Human-Robot Interaction Control Through Demonstration

In human-robot interaction (HRI) tasks, robots are required to change the behaviours according to different applications. As different interaction behaviours require different task requirements, it is difficult to formulate a HRI control problem in a general or unified way that enables robots to learn a set of task requirements or behaviours through human's demonstrations and then execute the tasks by using one controller. This paper aims to solve this problem by using a dynamic potential energy function to describe a set of different task requirements so that the motion behaviours demonstrated by human can be acquired or learned by the robot systems in a unified way. A control strategy is proposed to enable the robots to perform various tasks demonstrated by human and also change the behaviours during HRI according to different applications in a stable manner. The stability of overall system is shown using Lyapunov-like analysis and experimental results are presented to illustrate the performance of the proposed control strategy.