No abstract

Recently, robot scheduling has been widely concerned by numerous companies and researchers. They adopt an intelligent algorithm to design scheduling strategies for automated guided vehicle (AGV) and achieve required moving routes and certain aims. However, the existing robot scheduling strategies employ mathematical program models to arrange robots’ control and ignore machine issues that the robots may have. This will cause conflict and may lead the whole robot to a deadlock when the scale of conflict reaches a critical point. In this paper, we present a novel scheduling strategy for intelligent robots by training machine learning algorithms. Initially, we assume the robots contain a communication system, which can only receive signals from the central controller. Subsequently, we utilize the input parameters including robot conditions, working area, and working aims to train the machine learning algorithm. The work accomplishment ratio and conflict ratio are two essential indicators to measure our proposed method. Additionally, we also simulate the existing robot scheduling models to compare. From the extensive experimental results and comparison evaluations, we can conclude that our proposed method can achieve the robots scheduling tasks with reasonable costs. At last, we discuss the challenges of applying machine learning techniques to robotic scheduling and potential solutions and provide future directions in this field.