The Research On Motion Planning And Coordination For Robot Systems

With the development of computer, electronics, communication, control, sensor, and other related technologies, robotics develops rapidly and is applied to a wide range of domains. On the one hand, the researchers aim to design more complicated mechanisms, to mount various advanced sensors on robots, and to develop more efficient algorithms, which improve the intelligence and ability of the single robot to make it capable of performing successfully the tasks and adapting to more complex environments. On the other hand, multiple robots with relatively simple functions are employed to complete the tasks with complicated, uncertain, intercurrent, and real-time properties that cannot be done by a single robot.Before executing the given task by a single robot, path planning is necessary. It is the same case for a multi-robot system. Motion coordination among the robots is needed in order to avoid collision. It has been widely used for current motion planning and coordination methods in various tasks. However, they can only be exploited on special occasions and are heavily relevant to the tasks and system properties. If the tasks or environments are changed, these methods may not work well. Therefore, they lack of flexibility and generality. In addition, these approaches are able to plan and coordinate the robotic tasks on the implicit condition that the tasks are realizable, whereas the system capability as well as the solutions to unachievable tasks have seldom been addressed.In this dissertation, we propose a general motion planning and coordination strategy for robot systems, which offers a unified way for planning of single robot systems and coordination of multi-robot systems. It is available in different environments, tasks, and robot systems. The State Space for robot systems is constructed by choosing rationally state variables according to the task requirements and system characteristics. The internal physical constraints, inherent in the system, and external obstacle constraints and task constraints, imposed externally, are mapped into the unrealizable areas in the State Space. Then reachable areas of the system state denote the system ability to complete the tasks. Task execution is considered as transition of the system state in reachable State Space. Thus, the planning and coordination problem of the robot systems is translated into a trajectory solving problem in State Space.The realizable conditions for the tasks are given. If the system's initial state and the task's goal state are both in the reachable space, as well as there exists a connectable path in between within the reachable areas, the task is realizable. The optimal strategy for task fulfillment can be investigated and obtained according to the given performance index. If any condition above is not satisfied, the task is unrealizable. It could be transformed to be realizable by adjusting the system's configuration and/or task constraint, and the transformation condition for task realization could also be figured out. This lends itself to task designing, planning, and coordination.It is applied to planning and coordination of different robot systems and tasks in order to validate the effectiveness and generalization of the proposed method. In manipulator path planning tasks, the system ability and characteristic with/without obstacle constraints are analyzed. The shortest trajectory of the end-effector is obtained in the State Space for a Point-to-Point task. The unrealizable task is also addressed, as well as the conditions of critical configurations and critical constraints are derived. The state trajectory trackings are conducted both in simulation and experiment. Path planning for a mobile robot and a mobile manipulator are conducted. Real-time path planning and tracking are realized on an internet-based office robot in dynamic environment. The optimal trajectories of the mobile manipulator under different task requirements are obtained in State Space. For a multi-robot formation task, the minimal path transition trajectory of the formation is solved in State Space. The unrealizable task is addressed, and the constraint transformation contion for task realization is derived. The trajectory trackings are done both in simulation and experiment. In a multi-manipulator task of operating cooperatively a single object, the optimal solutions under different task requirements are obtained in different State Space. Then motion coordination between multiple manipulators is realized.