Design of protocol for collaborative task assignment in heterogeneous robotic material handling systems.

摘要

Heterogeneity in robot model mix is advantageous in emerging robotic material handling systems. For instance, in a fully automated industrial package loading and unloading scenario, a variety uncertain types of packages need to be continuously sorted and loaded onto designated trucks. To handle the variation of tasks, multiple types of robots, or heterogeneous robots, are designed in the system. However, in such a collaborative system consisting of heterogeneous robots, ineffective task assignments often lead to bad collaboration and thus poor efficiency. In this thesis, to improve the robot collaboration, the collaborative task assignment problem is defined; and a protocol with fuzzy collaborative intelligence to optimize the assignment plans is developed. Specifically, the collaboration type, the collaboration matrix and the assignment matrix are specified; and the model for adaptive fuzzy collaborative task assignment relies on intuitionistic fuzzy set theory. An unsorted package loading and unloading tasks by heterogeneous robots are used as a case example to validate the new method. Experiments indicate with statistical significance that the new approach shortens total completion time by 21%, reduces total energy consumption by 23%, and increases loading accuracy by 31%, compared with the traditional static task assignment method commonly practiced. The developed approach can be applied to different emerging collaborative systems to improve systems' collaborative intelligence.