Research On Planning Technology Of Multi-arm Space Robot For On-orbit Service

With the deepening of space research and the development of space technology,more and more spacecraft for various mission requirements have entered space.The structure and composition of the spacecraft have also become more complex,and the performance and technical level of spacecraft have continued to improve.However,at present,most spacecraft do not take into account the need for onorbit maintenance and upgrades during design,and once they fail,it will be difficult to recover.Therefore,how to enable spacecraft to operate stably and reliably in complex space environments and upgrade components in orbit is a research hotspot in the aerospace field.The use of space robots for autonomous on-orbit services shows great advantages and benefits,which can reduce the cost of space missions,extend the life of spacecraft and expand the functions of spacecraft.The autonomous on-orbit service requires a certain degree of intelligence of the space robot,including technologies such as autonomous task planning,trajectory planning,and robotic arm control.Compared with single-arm space robots,multi-arm space robots can perform more complex on-orbit service operations.Therefore,research on multi-arm space robot planning technology has broad application prospects.This article has conducted in-depth research on the planning technology of multi-arm space robot.The specific research contents are as follows:Firstly,for the situation that the space robot base is controlled,a path planning method for multirobot based on improved A* algorithm is designed.Different from the application scenarios of the traditional A* algorithm,it is adapted to meet the needs of real-time collision detection and multithreaded synchronous calculation.This method can avoid complicated free space calculations and shorten the path search time with more flexible search steps.Simulation results show that the improved A* algorithm can quickly obtain a safe and good path.Secondly,in the case of free-floating flight mode,taking the space robot to capture the target satellite as the application background,a dual-arm trajectory planning method based on reinforcement learning is proposed.Due to the dynamic coupling between the movement of the robotic arm and the attitude of the base in this flight mode,the positioning and orientation of the end effector will be greatly affected.In addition,the increase in the number of robotic arms has exacerbated the coupling effect and increases the complexity of the problem.Therefore,it is proposed to use reinforcement learning to learn the policy of this task.This method does not establish the kinematics and dynamics model of the space robot,but builds a task environment in the physics engine,and allows the space robot to learn the optimal policy through a large number of trial and error in this environment.Simulation results show that the method can quickly and accurately plan the trajectory of the robotic arm in real time for all situations in the capture range,and has certain robustness that can deal with the situation where the target and the service robot have relative movement.Finally,for typical on-orbit service operation tasks,the multi-arm task planning problem is studied.Utilize human experience and knowledge to perform task decomposition of on-orbit service tasks to obtain action sequences,and use the path and trajectory planning methods proposed in this paper to perform path/trajectory planning for actions in the entire operation process to achieve automatic system task assignment.It further proves that the method proposed in this paper has strong adaptability.