Research On Motion Planning For Close Proximity And Capture In On-Orbit Maintenance Mission

On-orbit maintenance has recently been a research hotspot in space monitoring and service.The autonomous spacecraft approach and the space manipulator capture are the essential foundations for the successful execution of on-orbit maintenance tasks.At present,the key to completing these two stages of autonomous operation lies in proper motion planning.Due to high dimensionality,strong constraints,and high requirements for optimality and real-time performance,it is incredibly challenging to design an applicable motion planning method for it.Therefore,this dissertation focuses on the motion planning strategy of spacecraft proximity and space manipulator capture.Aiming at the motion planning problem in the final proximity stage between the spacecraft and the space target,considering the constraints during the approach process of the spacecraft,a motion planner LG-RRT* based on the optimal rapidlyexploring random tree is designed to solve the approach motion planning problem.The feature of this motion planner is that it uses deterministic sampling to improve the original sampler to ensure the repeatability of its planning results,and its local planner is suitable for differential constraint systems.At the same time,the goalbiased strategy added in the planner makes the solution process more oriented.The performance analysis experiment results of LG-RRT* show that the improved strategies proposed in this paper are effective.The simulation results of LG-RRT* in spacecraft proximity motion planning prove that the planner is feasible.For the space manipulator to capture a space target with a stable motion state,considering that the long capture process will accumulate the impact of the manipulator movement on the base,which in turn will lead to the problem of erratic movement of the manipulator,this paper conducts the time-optimal trajectory planning of the space manipulator.Firstly,a parameterization technique is used to construct the capture trajectory,and a novel intelligent optimization algorithm(Egret Swarm Optimization Algorithm,ESO)is introduced to solve the trajectory parameters.Then improve the initialization population method and add an opposition-based learning mechanism to optimize the algorithm.Finally,the optimized algorithm HOESO is applied to the trajectory planning of the space manipulator.Simulation experiments prove that HOESO has more advantages than ESO in solving complex problems,and using HOESO for trajectory planning problems is feasible.A motion planning solution based on target pose prediction is proposed to aim at capturing the unstable space target by the space manipulator,considering the complexity of the motion state of the unstable target.The scheme uses a bidirectional long-short-term memory network as the basic structure of the motion prediction network.To reduce the impact of the complexity of the target motion state on the prediction network,a temporal attention layer is added to the basic structure to obtain the final prediction network model.Finally,based on predicting the motion pose of the unstable space target,the parametric trajectory construction method is used to plan the capture trajectory.The final simulation results show that the proposed motion planning scheme based on prediction is effective.