Research On Random Sampling Path Planning Of Ship Assembly/Disassembly With Constraints

The design of assembly/disassembly maintainability design is important to improve the efficiency of ship design, building and maintenance, as well as its performance and quality. In order to inspect the validity of design, it is essential to find a collision-free path for line replacement unit (LRU) in the virtual environment along which LRU could be moved freely into(out of) a part.But in the real applications, there are many difficult problems in finding the feasible path because of the complexity of the operating environment and the influence of all sorts of constraints. This paper studied several path planning algorithms of ship assembly/disassembly considering all sorts of constraints. First, this paper classified the constraints existed in path planning, then this paper studied sampling strategies, local planners and distance metrics of the motion planners considering the constraints to improve the efficiency of path planning.First, the paper proposed a new approach based on Rapidly-exploring Random Tree (RRT) algorithm considering rotation constraints. There is one problem in RRT algorithm that the random sampling function of rotating weights of Euler angles in3D-space is non-uniform, so a new random sampling function of Euler angles was presented which is uniform. Then, the Euler angle distance was introduced to distance metric function in configuration space (C-space) to control the variation of rotation angles and guide the direction of path planning. Experiments showed that the new algorithm could effectively constrain the rotation of LRU in high dimensional space, improve the computational efficiency and decrease the variation of rotation angles. Another new RRT-based algorithm including rotation constraints was also presented. In RRT algorithm, the transformation efficiency of rotation matrices during controlling rotation constraints is low, so the paper presented a new algorithm which utilized a new uniform sampling strategy based on unit quaternion and added a unit quaternion inner product into C-space distance metric function for controlling the variation of rotation angles. Considering motion constraints, a new random sampling-based path planning algorithm based on Bi-RRT algorithm was presented. The algorithm guided the samples along the constraint passage and restricted the extended RRT Tree nodes in the constraint volume. Experiments show that the algorithm much more effectively solved planning problems with motion constraints.The paper also proposed a random sampling-based path planning algorithm based on RRT algorithm with Gaussion strategy. The algorithm sampled more configuration points in important regions such as narrow passages so as to increase the connectivity of the free C-space. The results show that the algorithm could increase the speed of path planning.To solve assembly/disassembly path planning in those environments with dangerzones, a based-RRT path planning algorithm was presented. The algorithm addressed a random sampling function which obtained random configurations from the triangles that described the dangerzones. The local planner connecting the samples with RRT tree was also introduced. Experimental results show that the algorithm had better performance in searching paths between dangerzones and obstacles. In addition, the LRU should avoid dangerzones as far as possible, so an assembly/disassembly path planning algorithm considering environment danger level was presented. First, to ensure that the LRU does not collide with obstacles, repulsive forces were applied and artificial potential was built. Then the algorithm used the potential function to build danger coefficient that present danger level and constructed the heuristic function avoiding the dangerzones with the danger coefficient. Finally, the paper tested the validation of the algorithm on one cabin and the experiment results show that the algorithm had good performance in finding the path and could effectively avoid the dangerzones.Adding virtual mannequins in the process of part assembly simulation highly increases dimensions and difficulty of path planning. To solve the problem, the paper described a new technique based on RRT algorithm that could plan the path in high dimensions. First, incremental sampling strategy based on C-space decomposition was proposed, then a local planner suited for the C-space decomposition algorithm was built and the collision checking process between virtual mannequins and parts and between the articulations of virtual mannequins themselves was optimized. Finally, a complex environment with virtual mannequins was built to check the validation of the algorithm. Experiments show that the algorithm was much more suitable for the path planning with virtual mannequins.Finally, according to the algorithms proposed above, we designed an assembly/disassembly path planning system based on SDK library of CATIA software. We programmed with C++language and designed a path planning library that includes several randomized sampling algorithms proposed above. Then the explicit models, location data and assembly/disassembly information were integrated into the system designed by CAA. At this time, we combined the path planning library and the database, planned the paths in CATIA software and showed the simulation results.