Micro robot soccer game : motion control, path planning and game strategy

Abstract

This project is based on the Micro Robot Soccer Tournament (MiroSot), in which a wide range of technologies are integrated into a robot soccer game. Among these technologies, three areas are investigated: position and path-tracking control of the wheeled mobile robots (WMRs), the robot path planning and the game strategy. A WMR (the robot soccer player) is a highly nonlinear system with practically inevitable parameter uncertainties and disturbances. To drive a WMR to move faster and yet give a stable response, many researchers used the sliding mode control (SMC) method. In this thesis, we propose two control methods. The first method makes use of a heuristic fuzzy logic controller to control the WMRs. It is very simple so that it can be realized in the restrictive environment of MiroSot. The performance in path tracking of the proposed fuzzy logic controller is compared with that of two published controllers. Both software simulation and hardware experiment results show that the proposed controller can give a fast response. The second method makes use of the kinematic model of the WMR and its TS-fuzzy plant model to design the fuzzy controller. The stability of the fuzzy controlled WMR has been proven and the performance of the closed-loop system in tracking with a reference model is investigated. Simulation results will be given in this thesis. Next, two path-planning methods will be discussed. The first proposed method involves two types of actions of the WMR to form the path: angular movement and translational movement. These actions are executed one by one. Turning point(s) is/are set up based on the obstacle position(s). The final path is obtained by connecting the starting point, the turning point(s) and the target position together. This method takes advantage of the characteristics of MiroSot. Thanks to its simplicity, this algorithm demands only a low computational power. It gives a fast response upon detecting an obstacle, and takes the avoidance action within a very short time. This path-planning algorithm is tested in a self-developed robot soccer simulator, which provides a portable MiroSot environment for developing and testing various path-planning methods and game strategies. The second path-planning method is based on a grid-based approach. The stadium is divided into a number of small areas, and different values are assigned to them according to the robots' moving direction, the target point and/or the ball's movement. Each object is associated with a layer of small areas. The planned path is obtained by combining all layers and taking the minimum values. The ability of obstacle avoidance of this method is tested by the developed simulator. Two different game strategies have been designed. The first strategy is called Shortest Distance Strategy (SDS). Based on this strategy, the WMR having the shortest distance from the ball will take action. This method is used as a reference to test another proposed game strategy which is called Mode Dependent Strategy (MDS). According to the bail and robot locations, ore of the following three modes: attack, neutral and defense will be adopted in the MDS. Each mode is associated with a sequence of states that govern the WMR 's motion. This game strategy is put to tests of simulated robot soccer games. It is shown that the MDS performs better than the SDS.