| Target tracking is important for UAV research.When UAVs perform tracking tasks similar to target docking,capture,and landing,UAVs must meet spatial dimensions’constraints and reaching time constraints.Inspired by how animals guide and control their purposeful movements,this thesis is based on Tau theory and design UAV perception,planning,and control algorithms for tracking tasks with time dimensional constraints and smooth docking requirements.The main contributions of this thesis can be summarized as follows:(1)A Tau estimation method based on the priori knowledge about the target size is proposed.Given that the target type to be tracked is known and that the same type’s target size is in a similar interval,the relative distance is calculated according to the estimated target size.Then Kalman filtering is used to smooth the trajectory,and finally,Tau is calculated according to the filtered result.This method is less sensitive to the size prior error because the relative position and relative velocity errors caused by the prior error will be cancelled when calculating Tau.Furthermore,the calculation is simple,which can meet real-time requirements.(2)An improved Tau-G strategy is proposed,which generate spatiotemporal trajectories for any initial relative motion state.First,the current common Tau strategy and its limitations are analyzed,and then based on the characteristics of the initial relative motion in the tracking task,a falling body motion with an initial velocity is designed as a guided motion.The generated reference trajectory can meet the initial relative motion conditions and stable docking conditions.(3)For the improved Tau-G strategy,Tau controllers under different initial conditions are designed.According to the characteristics of different trajectory stages,a piecewise control rule is designed.In particular at the tracking stage,a non-linear P1D controller is used,which can mitigate the influence caused by the Tau scale change,then the controller is verified through experiments.(4)Simulation and real flight platforms are built.The simulation platform is built through the Gazebo simulator.The actual flight platform is based on the DJI MATRICE 100 and is equipped with an Intel NUC Onboard computer,Guidance system,and MYNT cameras.The simulation and real platforms’ software systems are built through ROS,which can complete data transmission between different modules.Finally,the effectiveness of Tau information in the UAV tracking systems is verified by experiments. |
