Path Replanning and Collision Avoidance in Collaborative Human-Robot Waste Sorting

Shortage of raw materials and expansion of waste around the world raise challenges in the field of recycling, as well as the automatization of recycling procedures. In the past decades, there has been an effort to deal with these concerns in a way that builds machinery that will automatically sort waste. This paper contributes to developing a more flexible and versatile station for mixed waste sorting. Including human-robot collaboration in the waste sorting process, both concerns related to safety and efficiency become prominent. To that end, this paper presents an algorithm for 3D human collision model generation from 2D images and the intrinsic matrix. Furthermore, the paper provides an implementation of an algorithm for replanning robot paths based on predicted collision from understanding dynamic scenes. The results demonstrate a concept and preliminary results of a robot path replanning algorithm based on a human collision model in a human-robot collaborative sorting.