In this thesis, the technique of design, modeling, simulation, off-line programming of graphics-based robot and its workcell, and that of cognitive control of behavior-based autonomous mobile robot are studied throughout. A series of valuable results are obtained on the algorithm design, software development, and experimental verification of the theory and method concerned. · The data structure and geometric modeling technique adopted in system design and development are discussed. We have implemented 3D geometric modeling of robot and its workcell in direct, indirect and interactive methods. The direct method means that modeling is realized through the programming way, which applies to modeling of basic objects or fixed workpieces. The indirect method is based on data through I/O interface produced by other CAD software. The interactive method means that users interactively create desired environment objects in the designed system. These three modeling methods are aimed at different purposes and different levels, including system development level and user level. · The forward kinematics and inverse kinematics of IRB2000 robot are discussed and simulated based on an coordinate system established for the robot, and several path planning approaches are given. Besides, a teaching system in the system soft environment is designed through which we can teach and program the robot in either graphics or off-line mode. Moreover, the process of the robot program execution can be displayed with the help of animation. · A program language and associate files ( e.g. data file, internal file and so on ) are designed, and a compiler and a decompiler are implemented. By the compiler, the source codes and the data file can be converted to the object codes acceptable by the robot controller, and the internal codes can be intepreted by the simulator; The object code in robot controller can be converted to the source codes and the internal file through the decompiler. · A approach to collision detection in the robot task execution is discussed, and the communication module is designed. The robot program produced in off-line mode can be downloaded into the robot controller and then controls the robot to carry out the desired tasks. The off-line program codes for several application examples of the industrial robot are given. · Behavior-based robot cognitive control is studied using reinforcement learning idea. In accordance with behavior-based robot cognitive control, its design ideas and implementation are discussed, including definition and implementation of sensors, definition of state space and action space, processing of different information and implementation of learning algorithms. For different kinds of robots and environments, the experiments with learning method and non learning method are made using either s
修改评论