Formation control of multiple robot systems with motion synchronization concept

Abstract

Study on coordination of multiple mobile robots has received increasing attention in recent years. One of the most challenging goals in robotics is the development of intelligent robots that behave as a team and perform tasks cooperatively with human-level performance, without requiring accurate knowledge of the kinematics and dynamics of team members, or an accurate model of the robot environment. It is a requirement that such a team of robots can maintain a certain formation during the course or change formation when needed. In this study, a synchronous coordination control approach is developed using cross-coupling control concept. This approach can be applied to swarms of mobile robots in switching between formations. The synchronization approach is extended to formation control applications, and the formation control problem will be posed as a motion synchronization problem. According to the desired formation, a synchronization control goal is derived, based on which the position synchronization error is defined as differential position errors between every pair of two neighboring robots. A mathematical model of generalized superellipse with varied parameters to present various types of formations is studied. Such shape regulation technology concerning switching between formations is developed to help apply the proposed control concept. Combined with this model, the proposed synchronization control strategy will be better supported and more effective. A decentralized cross-coupling controller for each robot is developed to stabilize its position tracking while synchronizing its motion with others for the desired formation. The control algorithm utilizes feedback of both position and synchronization errors, requires the information of the two neighboring robots only, and responds to all linked robots in the group. It is proven that the proposed controller can guarantee asymptotic convergence to zero of both position and synchronization errors. This controller does not require exact knowledge of the robot dynamic models and its environment and only requires motion information of its nearest neighboring robots for each robot control. To improve system robustness against uncertainty in controller design, an adaptive control approach is incorporated into the proposed synchronization control frame. The designed controller is verified to be robust to the robot model uncertainty and some external disturbances. There is always a chance that one or more robots will get stuck or damaged during tasks. A neural network approach is proposed to monitor formation controls of multirobot systems. A single neural network detector is developed to detect robot malfunction with data from robots’ odometers. Simulations and experiments have been conducted to demonstrate the effectiveness of the proposed approach. Both simulations and experiments are performed on switching tasks amongst different formations. The research outputs will benefit the widely used applications of multirobot systems in modern lives, such as search and rescue, surveillance, transportation, etc.