Research On Dynamic Path Plan For Agent Robot

Path plan is an important topic in mobile robot field. For the information got by sensors is limited and using this information, the robot's actions should accord with robot's kinetic restriction, mobile robot dynamic path plan is always a difficult problem in path plan field. In this thesis, robot system structure and path plan algorithm are studied to realize the mobile robot's dynamic path plan. At first, we modularize the robot's tasks, for instance, orientation, sensation, communication, action, and based agent conception, we realize the agent robot system structure on P2-Dxe robot from ActivMedia company using multi-thread technology and multi-protocol communication method. Next, the action module is analyzed and designed. In view of the limitation of information and satisfaction of fast response, a layered control system is adopted. The control system partitions robot's tasks into four parts those can restrict with each other. Moreover we use 2-D laser measure system to fulfill robot's environment modeling, and average the sensor data to improve measure precision. Artificial potential field is the main algorithm in this thesis. After selecting the right potential function, the robot's mathematic model is combined with the relevant potential force. Effected by the potential force, during analyzing the robot's kinetic and kinematic functions, robot's action is made. Finally, we experiment the agent robot system and the dynamic path plan algorithm, and analyze the robot's velocity and rotation velocity. The experimental results reveal the availability of the agent robot system and the path plan algorithm.