Human intention and workspace recognition for collaborative assembly

摘要

Human-Robot-Collaboration is being used more often nowadays in assembly processes that can not be fully automated in an optimal way. Skill-based task sharing showed to be a good alternative to manual labor mainly when tasks are either non ergonomic or require high precision. While collaborating with humans, robots are usually equipped with sensors to perceive the environment or the human when necessary. This paper presents a method for not only recognizing the product being assembled along with its components but also tracking the live motions of the human hands in the assembly area. This method can either be used as an input for the decision making module of the robot, or as a measure for quality insurance and human fault detection.