Research On Sampling Strategy Of Lunar Rover On-board Manipulator

Deep space exploration becomes an important development direction of human's space activity for it plays an important role in human's research of the formation and evolution of cosmos.The moon becomes the primary goal of human's study on deep space because of its unique geological position and potential resources.Selenographic exploration mainly contains using lunar rover to survey moon's surface environment and collect sample.With the increasing development of selenographic research,the existing selenographic sampling technology cannot satisfy the demand of increasingly complex sampling task.Therefore,the research of safer and higher-efficiency selenographic technology is significant to speed up the lunar research.In this paper,based on the cooperation project with the China Academy of Space Technology named"Intelligent Strategy Analysis Software for Sampling and Packaging of Lunar Soil",the research of the relevant strategy of sampling task executed by lunar rover on-board manipulator is studied,and the simulation and experiment of the research result is carried out to prove its feasibility.A learning planning algorithm based on knowledge base is put forward in this paper to solve the problem of inefficiency in the process of repeat sampling task execution.Firstly,the selonographic sampling task is decomposed into atomic task which can be recognized and executed by manipulator directly.Then,a modified PRM algorithm is put forward to solve the problem of inefficiency in the process of a single sampling task execution.The knowledge planning base,which can extract the feature from past planning result and save it,is introduced into PRM to enable the algorithm to get experience from historical planning result,which improved the planning efficiency greatly.Considering the situation of contact and collision between the end-effector of on-board manipulator and sample soil in the process of sampling,the optimal design of end-effector sampling scenario is carried out.Firstly,the dynamic model of on-board manipulator is established based on selonographic environment.Then,the contact mechanical model between end-effector and soil sample is built based on the trait of lunar soil and the structure of end-effector.Then the sampling scenario of minimum contact force can be achieved by analyzing the relations of contact force and sampling angle and sampling velocity.Finally,the position and force control algorithm is designed based on the dynamic model of manipulator and contact force model of end-effector and sample.A sampling contact force control strategy is put forward to ensure the manipulator can take measures in time once the rigid collision happened,so that the sampling task can be executed safely and effectively.The sampling configuration optimization of on-board manipulator is carried out to solve the problem of resource consumption in the process of sampling.Firstly,the sampling circle is generated by generalizing the sampling point,which improved the flexibility and selectivity of sampling scenario.Then the sampling configuration of manipulator on different sampling position can be achieved by analyzing the optimal sampling scenario.Then the motion pattern of manipulator can be achieved by combining the PRM algorithm based on knowledge base and planning.Finally,the multi-objective optimization model is established based on task demands,and the optimal sampling configuration can be calculated based on multi-objective swarm optimization algorithm.To prove the effectiveness and feasibility of the learning planning algorithm that is put forward in this paper,the experimental research of selenographic sampling task is carried out.The research result we get from the experiment is consistent with the expected goal,which proved the feasibility of the algorithm that is put forward in this paper.