Performance Evaluation of Intention Recognition in Human-Robot Collaborative Environments

摘要

Robots are becoming more and more prevalent in our everyday lives. They serve many roles, including acting as autonomous military scout vehicles, locating and destroying hazardous objects, welding in manufacturing plants, and vacuuming homes. However, in almost all cases, these robotic technologies have to be physically separated from humans due to safety concerns. This is primarily because of robots' lack of ability to perceive and understand their environment and project or infer what will happen in the future. While humans often do this naturally, it is currently a significant challenge for robots. This projection is often referred to as "intention recognition." In this article, we propose a novel approach to intention recognition based on the recognition of states in the environment. We also propose a set of metrics that can be used to assess the performance of this type of intention-recognition system, and we then apply the approach in a modeling and simulation environment.