Study On Behavior-Based Control Of The Mobile Robot In Unknown Environment

Supported by the national 863 project of China (No. 2003AA735084),this thesis deeply studied and discussed the problem of behavior-based controlfor the lunar rover.Lunar rover is one of mobile robots for exploration of unstructuredenvironment. As the forthgoer for the astronaut's landing, the lunar rover isvery important for the moon exploration and research. Therefore, lots of timeand money is spent by many developed countries in designing their ownplanet rovers.As matter of fact, the lunar rover usually moves on unstructured surfaceon the moon, thus, this thesis emphatically studies design of basic behaviorand strategy of motion planning based on the behavioral control viewpointand the fuzzy control method applied. And hence, two navigation methods ofarbitration mechanism and fusion mechanism are proposed followed bysimulation results which prove the validity of presented algorithm asbehavioral control structure is considered. Then, to integrate all of theprograms in this algorithm and realize the fundamental man-machineconversation the GUI interface is designed in the MATLAB environment.This thesis is organized as follows:Firstly, the background of the problem and the development of the moonexploration in the world are introduced. The relevant literature about thecontrol algorithm of the path planning and motion planning for mobile robotsare summarized, especially the behavior-based control methods.Secondly, the behavior-based control algorithm is studied deeply. Thebehavior-based hybrid control system is designed for the reason that the lunarrover has to move in the unstructured environment. Furthermore, theconception of the lunar rover agent is applied and the complex environment issimplified to 2-D model, which ground a good foundation for validating theefficiency of this algorithm.Thirdly, the basic behaviors such as the goal-trending behavior,obstacle-avoiding behavior and edge-following behavior are designed byusing the fuzzy control method to avoid the bad real-time performance of themodel-based method. As we know the global map information can not be get,so the reactivity is very important for the safe of the lunar rover. For the abovereason, the inputs of the fuzzy controller are the radar detecting data and thedistance between the goal and the lunar rover for different behaviors. Theoutputs of the fuzzy controller are the steering velocity and the translationalvelocity of the lunar rover. The design of the agent of lunar rover is based onthe Pioneer-Ⅱ mobile robot, hence this agent-based method can be applied toother mobile robots undoubtedly, as well as the lunar rover. In addition, thesimulation results based on the 2-D map proceeds to prove the validity of thealgorithm.Then, the sensor-based method is proposed for the real time obstacleavoiding and motion planning to guarantee the reactivity of the lunar roveragent moving in the unknown environment. Moreover, the navigation strategyis designed using the arbitration mechanism and fusion mechanism. Hybridarchitecture based on the layers of control system is built through thearbitration mechanism. In this architecture, "planning" is designed as one ofthe behavioral layers. It decides which behavior is activated according to thedata of sensor. Through the fusion mechanism the different weighting factorsare added on the different behavior and they are tuned by fuzzy controller withradar detecting data. Weighting factors as the crisp out of the controller aretaken into account to compute the final motion commands steering velocityand translational velocity. After that the evaluation measures based on theenergy (the ability that robot can save energy and time while moving if thetrajectory is smooth and short) are proposed to test respective performanceindex under different control architecture.Finally, regarding to the integration of M-function files and conveniencefor direct simulation, fundamental man-machine conversation interface isdesigned using the GUI tools of the MATLAB software. In the end, the wholearticle is summarized and the direction of the latter work is indicated.