The Study On Multi-agent Triangular Formation Control With The Angle Gradient Control Law And Its Experimental System

Agent is a model of knowledge processing system, which could be regarded as an individual with purpose, behavior and perception. Multi-agent formation control is to make more than one autonomous agents with simple detection capability keep the shape of a formation of desired. The formation control problem is the study of control algorithms of the independence movement of the agent in order to ultimately achieve the formation of a purpose of the external geometry. The essence of this research is the collaboration of multi-agent task or group of intelligent problem. Multi-agent triangle formation control problems are studied in this paper, including the proposing of control algorithm and the development of experimental platform.1. Put forward the multi-agent triangle formation control algorithm. The paper analyzes the study of multi-agent formation control problems, defects of the gradient control law in a triangle formation control problem and the proposed control algorithm is based on the angle of gradient. The angle gradient-like control law is used in a triangle formation control and resolves the defects of gradient-like control law. There is an in-depth research on the use of the angle gradient-like control laws to achieve a nonlinear dynamical system posed by the acyclic triangle formation. The paper studies the equilibrium of the system and proves the convergence and stability of the system.2. Complete the simulation of the agent triangle formation control under the angle gradient control law. In this paper, simulation platform of the agent triangular formation is set up in LabVIEW8.5, in which the agent triangle formation movement simulation is accomplished. We also study the process of the multi-agent triangle formation control and verified the angle gradient-like control law.3. Develop experimental platform for the multi-agent triangle formation control. This experimental platform is based on centralized control for multi-agent system on PC. The experimental models are three independent designed smart cars. Use a microcontroller as the core of the smart car’s control system, drive the model to move. Detection and control is realized by PC via USB Camera, real-time acquisition and feedback of the position and orientation of the smart cars, and real-time data processing. Use RS232as data exchange interface, realize the data exchange for host PC and lower microcontroller.4. Design image-based position and orientation detection system of smart car. Using computer vision and image processing technology, this paper develops a position and orientation detection system of smart car, also solves the problem of tagging smart car body. Bases on Vision module of LabVIEW software, image processing techniques of2-D images collected from USB Camera, including distortion correction, threshold segmentation, data calibration etc., are applied to achieve position and orientation detection of the smart car.