ROS-based Planning And Motion Simulation Of Mobile Manipulator

As a kind of novel robots, mobile manipulator plays a more and more important role in domestic service. More and more researches around mobile manipulator have been carried out. This dissertation presents a thorough research on the simulation platform and motion planning of mobile manipulator, and also puts forward some views and methods about the robot.The first part of this dissertation focused on the kinematics analysis of the redundant manipulator. It’s a simple task to build link coordinate system and solve forward kinematics problem while inverse kinematics issue is more complicated than that of common 6 DOF manipulators. In this part, the notation called arm triangular plane is presented, and then we can obtain current position of each joint by fixing the revolute angle of the plane. The correctness and availability of the inverse kinematics method are verified by line interpolation, circular arcs interpolation and self-motion interpolation in Cartesian space.Mobile manipulator is a complex system that consists of mechan ism arm, mobile platform, sensors, etc. Thus it’s more difficult to simulate mobile manipulators than traditional robots. This dissertation focused on how to create robot simulation platform based on the Robot Operating System(ROS). By making use of the open-source dynamic simulator named Gazebo, it’s simple to simulate Cartesian interpolation motion. The satisfying results prove the validity and reliability of the mobile manipulator simulation platform.Motion planning is the primary problem of robots’ moving. When offering services to human, mobile manipulator may encounter dynamic obstacles, in which case, motion planning will become more challenging. In this dissertation, we improve the traditional Rapidly-exploring Random Tree(RRT) algorithm and apply it to the dynamic environment. The ability of motion planning among dynamic obstacles has been significantly enhanced.In the part of virtual experiments, we complete two virtual experiments: grasping cup and opening door based on Kinect depth sensor. The last part of this dissertation focuses on simulating the motion planned by dynamic RRT algorithm on the ROS and Gazebo simulator. By testing the whole planning process of planning and controlling, the availability of dynamic RRT algorithm is verified.