Motion Control And Path Planning For Millimeter Size Mobile Microrobot Based Microassembly System

With the technology development of mobile microrobot, various performances of the mobile microrobot are improved greatly, especially on the mobility and motion precision. Relatively, the mobile microrobot based application researches become warm and warm. And much attention is paid on the micro parts assembly using mobile microrobot.Supported by research programs of China, our research team designs the millimeter size omnidirectional mobile microrobots, and develops the microrobot based microassembly system. The exploratory research on the micro parts microassembly with this system has carried on. For a microassembly system, the microassembly accuracy and precision are important performance indices. At present, there are some drawbacks on the aspects of microrobot motion control and microassembly path planning method, which influences the microassembly accuracy and precision. This paper presents the microrobot motion control and microassembly path planning of the microassembly system. The research objective is to improve the microassembly accuracy and precision, and to lay a foundation for realizing the precise microassembly operation.The research on motion control is the foundation of microassembly path planning research. The motion control research includes the wheel structure size design, motion characteristic, slip and stable step movement. The path planning research includes the intelligent path searching method, and partitioning of microrobot state space and path planning knowledge database.The dynamic analysis derives the function relationship between the needed micromotor torque for movement driving and the wheel structure sizes. Therefore, the genetic algorithm (GA) is employed to design the wheel structure sizes. The target function of GA is the relation equation of the torque and wheel sizes, and the optimal target is to minimize the torque consumption. The experiments show that the microrobots have good driving ability, which provides the guarantee of motion control.This paper presents more comprehensive research on the microrobot kinematics. Considering the microrobot special structure, the kinematic constraint matrix is derived and employed to analyze the maneuverability and omnidirectional characteristic. Besides, the kinematic equations of the microrobot chassis and microassembly arm are found. And the Jacobian matrixs are derived and used to analyze the microrobot and arm motion characteristics, and the analyzed results are same as the real? Supported by National Natural Science Foundation of China (No. 60475037), and Specialized Research Fund for the Doctoral Program of Higher Education of China (No.20060248057), Hi-Tech Research and Development Program of China (No. 2007AA04Z340). situation.The microrobot slip dynamic model is found. According to the microrobot structure characteristic, three factors which can generate slip are analyzed, and their slip effects are also analyzed and demonstrated based on the slip dynamic model. A vision based slip control method is developed to reduce the motion deviation.The micromotor step positioning principle based on torque self balance is analyzed, and then the torque expression for microrobot step movement driving is drived. The simulation model of step movement is found based on the microrobot dynamic equation. With this model, the step movement process is simulated. The simulation shows that the micromotor torque has great influence on the step movement stability. Therefore, the feedback control of micromotor current is used to obtain stable step movement and uniform step displacement.There is the problem of motion path selection during the microassembly process. The conventional path planning method can not ensure high microassembly accuracy and precision. Therefore, a mixed learning method that integrates the supervised learning and the reinforcement learning is developed and then employed for the microassembly path planning. Compared with the conventional method and the single reinforcement learning, the mixed learning method achieves higher microassembly accuracy and precision. Three models, named uniform grid tile, uneven grid tile and cobweb tile, are designed to partition the microrobot state space. Their application effects are compared, which show that the learning efficiencies of the later two models are much higher than that of the first model. The performance of the cobweb tile is best. Finally, the path planning knowledge database is found to improve the on-line path planning speed.Above all, according to the millimeter size microrobot based microassembly system, this paper presents the microrobot motion control and microassembly path planning. Some problems of them are analyzed, and relative solutions are developed and then demonstrated by simulations or experiments. Though some achievements are obtained, deep research is needed on some directions, e.g. multi microrobots cooperation for microassembly.