Human-Robot Collaboration System Based on Task Estimation Using Spatial Memory in Intelligent Space and Robot Task Selection by Markov Decision Process

Abstract

In collaborative work between humans and robots, we developed a system that can recognize a worker's work situation through spatial intelligence and respond flexibly and proactively. We evaluated the usefulness of this system through comparative experiments. The results confirmed that the proposed system enables the robot to select a task and execute an action at a time that considers the worker's work. However, it was suggested that more flexible task selection is needed, and a new robot task selection algorithm using a Markov decision process was developed. As a result of its operation verification, we confirmed that the robot could perform task selection more robustly.