Autonomous Navigation And Control Of Micro Multi-rotors Aerial Vehicle Towards Complex Environment

Thanks to the increasing complexity of the environment and diversity of tasks,autonomous navigation of Micro Aerial Vehicle(MAV)towards the complex environ-ment is attracted attention in the academic.To improve the capability and performance of MAV towards unknown complex environments,the problems of motion planning and control are considered in thesis.The main content of the thesis is as follows:Firstly,considering the autonomous flight towards the complex environment,a completed solution about the navigation of MAV is proposed.The measurements of states are generated through onboard sensors.The state estimation system is proposed to fuse the multiple measurements.Then,the map information for motion planning is constructed from the 3D point clouds.The motion planning system employs a global-local planning strategy.The global planner searches for a global initial path.The col-lision avoidance and dynamic feasibility are considered in the local motion planning problem.The local optimal reference trajectories can be solved by the nonlinear opti-mization method.Experimental testing reveals that the proposed solution can achieve fast and robust MAV autonomous navigation in complex environments.Secondly,the motion planning of MAV with limited perception is considered.The motion and perception objectives are integrated in one nonlinear objective function and formulated as a stochastic optimal problem.The proposed sampling-based Model Pre-dictive Control framework allows the motion planning problem to be solved in real-time.This approach is able to generate reference motions that are dynamically feasible and collision-free with guaranteed perception.Then,the method is also implemented for two challenging perception-limited scenarios.The effectiveness of the proposed method is demonstrated through simulation and experiments.Thirdly,a parallel and gradient-free strategy is presented for autonomous navi-gation with a limited field of view.The proposed method constructs an environment map by integrating the measurement of a limited-FOV sensor in a parallel process for improving the real-time performance and computational effectiveness.Compared to existing methods,its computational complexity is irrelevant to the resolution of the depth image and the number of perceived obstacles.The feasible initial resolution is selected from the global path to guide the local planner.The motion planning problem with nonlinear perception constraints is solved using a parallel-gradient-free algorithm.The translation and yaw movement can be achieved simultaneously by solving this op-timization.The proposed strategy is successfully implemented on a small-sized MAV.Real-time performance is achieved with the limited onboard computational budget.Its performance is demonstrated with real flight experiments in unknown indoor complex environments.Finally,considering the effect of external disturbance,the classical in-out loop control framework is implemented to improve the trajectory tracking performance.The trajectory tracking controller calculates the reference attitude through the tracking er-ror of position and velocity states.Based on the super-twisting sliding mode control algorithm,the multivariable finite-time attitude controller is proposed with fast con-vergence,high accuracy,and robustness.The effectiveness of the proposed trajectory tracking controller is confirmed by simulations and real-flight experiments.