Dynamic Analysis And Control For Multi-rotors Grasping

The rapid development of multi-rotors extends the applications of robots from ground to air, which has brought great changes to human life. At present, UAVs are largely applied to environmental monitoring and aerial recording by equipping sensors like cameras. However, the interaction between UAVs and objects in the environments is lacking in these applications. The UAV with robotic arm can manipulating surrounding objects, which can greatly expand the uses of UAVs. The grasping of multi-rotors distinguishes from the control of traditional UAVs. The main difficulty lies in the inter-coupling of dynamics between the UAV and the arm on it which makes the system model more complicated. So it is essential to generate accurate flight paths according to the actual pose of the target obje ct. But in terms of the UAV combined system with high dimensio Fnal states, traditional trajectory generation methods such as A* or motion sample-based RRT algorithm cannot be used normally, so there need some new methods to solve it.In this dissertation, a 3-D dynamics model for the coupling system of the quadrotors with a mechanical arm was built firstly, while the grasping model for UAVs established by the GRASP lab of Pennsylvania State University was a special case of the one presented in this work. Next, to solve the problem that traditional planning methods are difficult to use in multi-dimensional states, we designed a novel trajectory planner based on the built model by applying i LQR algorithm. The algorithm relies on the nonlinear dynamics model of the combined system to construct a non-quadratic cost function. Through linearizing the dynamics model and quadratizing the cost function, a 3-D motion desired trajectory for the system could be planned. The common used method called minimum snap needs to utilize differential flatness and does not employ system dynamic model. After generating the trajectory, a tracking controller was also designed based on the characteristics of the UAV and the arm to track the desired trajectory. The controller of UAV was structured by considering PID, geometric space method, and acceleration feedforward control method. The system was controlled by the inner and outer loop to pursuit the thrust and torque for driving the UAV. For the mechanical arm control, we simply applied a PID controller to control its rotation.To verify the novel approach proposed in this dissertation, we improved the Rotors-simulator, a UAVs simulation system based on Gazebo in ROS. By employing advanced dynamics engine OED, the improved platform is sufficiently close to the actual dynamics characteristics of UAVs. At last, the validity and superiority of the proposed approach was verified through the simulations. Apart from applying to the generation and control for the UAV-arm system grasping, the presented method and simulation platform in this dissertation can be extended to more research fields such as mobile manipulators, space manipulators, suspended load manipulation using UAVs and so on. It is a significant reference for other researchers. Furthermore, the work has important significance for our country to master core technologies of the motion planning and tracking control for the UAV autonomous system.