Research On Multi-robot Task Assignment And Path Planning For Space Truss Assembly

With the development of space missions,the research direction of in-orbit assembly technology of large space truss has changed from manual assembly of astronauts to autonomous assembly of robots.Research on the technologies of self-assembly,service maintenance and renewal of large space trusses can prolong their life cycle and reduce the risk and cost of future missions.China will soon enter the era of space station in 2022,and the in-orbit assembly technology of large space truss will play a huge role in promoting the development of China's space technology.For efficiency and safety reasons,robots are better suited to assembling space trusses than astronauts.Compared with the traditional single robot,the highly automated multi-robot system has the advantages of high reliability and high efficiency.In the face of long-distance transportation requirements,complex assembly environment and task constraints in large truss assembly tasks,the assembly path planning and assembly sequence allocation of multi-robot system must be carried out before the assembly task of in-orbit truss is performed,so as to improve the efficiency and safety of multi-robot in-orbit assembly.In this paper,spatial multi-robot system is taken as the research object,and the key technologies such as mathematical modeling of spatial truss assembly task,multi-robot assembly path planning strategy,multi-robot truss assembly sequence allocation strategy and so on are studied.The main research work is as follows:Firstly,the task design of space truss assembly and the task mathematical model construction are studied.By analyzing the task characteristics of space truss assembly and analyzing the task assembly process,a general multi-robot multi-objective point path planning task solving framework and multi-robot truss assembly sequence assignment task solving framework are established.By analyzing the principle of truss construction,the connection relation of truss and truss and the sequence relation of modules are established.By analyzing the task state and constraint condition of truss assembly,the task optimization objective function is established,and the mathematical model of space truss assembly task is established.Secondly,research on multi-robot assembly path planning strategy is carried out.Multiple robot in this paper to establish the target path planning task solving framework,on the basis of multi-robot path planning tasks derived the relationship between objects and constraints in the goal,overall optimization goal design task,and the algorithm based on self-organizing map network model multi-robot assembly path planning task for the establishment of the solution space.Considering the requirements of task solving speed and robustness,the target locking strategy is designed to lock the tasks in the competition process,so as to avoid redundant computation caused by repeated activation of neurons by multiple tasks,and to improve the problem of increased space complexity caused by task conflict.At the same time,the map space is segmented,the update direction of the shortest obstacle avoidance path of the neuron is determined according to the direction of the vertex line in the space domain,the process of obstacle avoidance path generation is accelerated,and the time complexity of the algorithm is reduced.On this basis,a new winner election strategy is introduced to solve the problem of multi-robot path conflict,so that multiple paths are not intersected,collision avoidance between multi-robots is realized,and the quality of algorithm planning solution is improved.Thirdly,the allocation strategy of multi-robot truss assembly sequence is studied.Firstly,based on the state and constraints of the multi-robot truss assembly task,the sequence allocation state transfer function was established on the basis of ant colony algorithm model,and the parameter optimization was carried out for the function solution.On this basis,the multi-robot truss assembly sequence allocation algorithm was studied.Then,consider the assembling error of task allocation under the risk of failure,the sequence of assembling error below state evaluation function,and based on this design error sequence adjustment strategy,with the error information to compensation adjustment of truss task state matrix,and the part of the sequence errors to optimization allocation,raise the stability of the algorithm under assembly error.Finally,the experimental study of multi-robot truss assembly task is carried out.Based on the requirements of the multi-robot experiment task,the experiment platform was reconstructed and the experiment scheme was designed.The validity of the relevant theory of multi-robot truss assembly proposed in this paper was verified and analyzed by collecting and comparing the experimental data.