Research On Methods Of Assembly Path Planning For High-speed Train Carriages

Assembly path planning is one of the foremost tasks in the virtual assembly simulation for the assembly process. The core mission of the assembly path planning is to find a collision- free path for parts to follow with from the starting state to the terminal state. A large amount of parts, complex structure, and compact interior space are characteristics of the high-speed train, and these characteristics make the path planning process difficult. In addition, as the complexity of the assembly environment and the influence of different sorts of constraints, there are many problems and challenges in the assembly path planning tasks for high-speed train.Firstly, this paper classified the constraints exist in path planning,and discussed the path quality evaluation methods. Then, this paper analyzed the traits and problems in assembly path planning tasks for high-speed train, and studied the key technologies for solving the problems.A reasonable solution space for path planning tasks is beneficial to improve planning efficiency and path quality. To deal with the problem of how to get a reasonable solution space, a method based on the information of plant configuration and assembly process was proposed to rapidly obtain a rough solution space, and a method was presented to calculate a more refined solution space. Besides, considering the actual start state of assembly path, a segmented assembly path planning theory was proposed. This theory provided an idea that the assembly path planning process could be divided into two stages. The traits of the two planning stages and the method of divid ing the two stages were also studied.The paper presented a double rotation constraints approach based on unit quaternion and rapidly-exploring tree(RRT) algorithm to solve the rotation constraints problems in assembly path planning. This approach imposed rotation restrictions in the procedure of generating a random selected state and a new state, which could control the variation of rotation angles between two states in a path planning task with a large rotation angle between the start state and the terminal state. Experimental results showed that the new approach could effectively constrain the rotation of parts. In addition, a new RRT variant algorithm including rotation constraints was presented to obtain minimum cost path solutions. This algorithm utilized a new uniform sampling strategy and a new extending strategy; the method produced several random states in uniform sampling procedure and in extending procedure. The tree extended toward the new state with the least cost which involved in translation cost and rotation cost. Experimental results showed that the new approach could efficiently improve the path quality.As the path planning process is difficult in a n assembly environment with narrow space, the paper proposed a new triple RRT method. In order to increase the connectivity of the narrow space and speed up the growth of trees, this method chose the tree which grows slowestly as the extending tree with a certain probability, then extended the tree with several step length toward random state. In connecting strate gy, connecting tree grew toward the nearest node in extending tree. This connecting strategy could improve the phenomenon of path retracing. Experimental results showed that the new triple RRT method much more effectively in solving path planning problems in an environment with narrow space.Considering path planning for virtual mannequins in the process of parts assembly simulation, a new biased sampling method based on C-space decomposition was proposed. This method translated the sampling procedure of mannequin posture into the sampling procedure of joint rotation angles. Three biased sampling factors were introduced for guiding the sampling procedure; one factor guided the sampling strategy choose the terminal state as the random state, one factor restricted the sampling strategy work for several joints, and another factor restricted the sampling strategy work within the comfort range. Furthermore, respectively from the path cost perspective and from the comfortable posture perspective, two evaluation methods for virtual mannequin path were presented. The simulation results demonstrated that the new biased sampling method was effective in speeding up the planning process with virtual mannequins, reducing the path cost, and improving mannequins' comfort.