On-Line Deadlock-Free Planning of N-Industrial Arms With Independent Controllers Using Advanced Escaping Method

Abstract

This work presents an on-line deadlock avoidance system for N-industrial-robot arms utilizing an advanced escaping method. Robots within the same workspace are controlled via independent controllers using point-to-point commands. Besides, the robots have no preceding information about the commands that will be sent to the controllers after starting the system of robots up. In practice, the deadlock situations, in which the robot becomes an obstacle in front of another in the collision avoidance process, are an industrially common problem in on-line planning. Therefore, to make the proposed on-line collision avoidance system more functional in industrial applications, a deadlock avoidance method is integrated into the system. We have previously proposed a simple escaping method to avoid the deadlocks among the end-effectors of two robots, thereafter, an escaping method for the whole bodies of two robots has been proposed. In this work, an advanced escaping method using our previously devised advanced collision map is developed to avoid the deadlocks of N-industrial-robot arms. The advanced collision map has been designed to detect potential collisions between all body parts of robots and to represent collisions as collision areas on the map. The map is created for the robot that has acquired the command and is going to execute it and other robots in the workspace. Thus, the robot treats other robots as either static or dynamic obstacles. Hence, for generating a collision-free trajectory of the robots, time scheduling of command execution time is applied to avoid any collision areas on the map. The effectiveness of the proposed collision and deadlock avoidance system is demonstrated by testing the system on an OpenGL-based simulator. In addition, the system is evaluated by analyzing the results of many simulations with different robot arrangement patterns in the workspace as well as by comparing the system with the most relevant work to this article.