Study On Motion Planning Of Simultaneous Mobile And Manipulation Of Robot

Mobile manipulation combines the mobile robot and the manipulator,which can extend the scope of operations effectively and improve the flexibility of manipulation.It becomes the necessary skill of a variety of service robots,new industrial robots,military robots and space robots.Mobile manipulation is very different from stationary manipulation.Fistly,the degrees of freedom become bigger when adding the mobile base.Secondly,the dynamic equation is changed after adding the mobile base,the trajectory planning becomes more complex.Thirdly,as always with task constraints,the collision avoidance of mobile manipulation is more complex than the collision avoidance of stationary manipulation.Existing planning methods of mobile manipulation always separate bases from robotic arms,which can lead to wasted time and some task constraints that require coupling bases to robotic arms.Combined with the actual task requirements,we have made a systemic and deep research about the kinematics and dynamics of mobile manipulators,and get a series of application effects.The research contents are listed as follows:(1)The whole visual servo motion planning of mobile manipulation is researched.Based on the differential of the kinematic equation of mobile manipulation,the global Jacobian matrixes of arbitrary points in the mobile manipulator are derived when the base moves on 2D planer and 3D space.And a new hybrid visual servo method(HVS)of mobile manipulation is derived based on the analysis of the strengths and weaknesses of the classical image based visual servo method(IBVS)and the classical position based visual servo method(PBVS).A new velocities control rule of mobile manipulation is derived combined with the hybrid visual servo method and the global Jacobian matrix.Using this velocities control rule,combined with Kalman filter,the mobile manipulator can make grasping while walking.A Lyapunov function is constructed to prove the stability of the HVS method we proposed.The mobile manipulator of this chapter which consists of a Bulldog chassis,an UR5 robotic arm,a Robotiq-85 gripper and a ZED camera help us make the algorithm come true.The results show that our method has the advantages of faster convergence to the object.(2)The path and trajectory planning of mobile manipulation are researched in complex environment.firstly,this paper deals with the evaluations of success rates of different grasp positions and orientations.Based on the success rates,the iterative optimization mobile grasp planning method is proposed.This method can not only optimize the position and orientation of grasp but also optimize the position of the robot base and the configuration of the robotic arm while grasping.With this method,the robot can reach an optimal configuration which benefits the motion after grasping.Then A*algorithm is used to plan the path of the base that connect the origin and the target position.Based on the path,we propose a heuristic random path approaching algorithm which is used to plan the motion of the arm so that the robot can walk and grasp at the same time.The self-made mobile manipulator which consists of a differential base,a Kinova-KG3 robotic arm and a Kinect camera help us make the algorithm come true.This paper utilizes the global Jacobian matrixes to deduce the dynamic equation of the mobile manipulator,and plans the trajectory based on a given path with using the second order centrum planning of the convex optimization.This method is verified in ROS(Robot Operating System)Indigo and Sedumi tool in matlab.(3)This dissertation studies the dynamical obstacle avoidance of arm end fixed mobile manipulation.We proposed a model predictive control method to plan the dynamical obstacle avoidance while mobile robot is manipulating the object.An object function which is designed to maximize the closest distance between obstacles and the robot and simultaneously minimize the velocities of null-space is proposed.And in order to optimize the function,using the global Jacobian matrixes deduced in chapter 2,we have deduced the model with 4 differential equations which represent the law of the distance over time(not the displacement over time).The model is consistent with the object function.And ACADO which is an open source toolkit is used to plan in real-time and get the smoother trajectory with smaller acceleration.The experiment is realized in Gazebo 2.0 simulator in ROS Indigo system.The self-made mobile manipulator of this chapter consists of a mecanum base,an UR10 robotic arm and 4 Kinect cameras.