A GA path planner based on domain knowledge for AUV

Autonomous Underwater Vehicle (AUV) is extensively used for marine engineering, oceanography research and some other civil area. Path planning is a necessary and fundamental technology for AUV autonomy, its goal is to search an optimized path according to some optimization criteria in a certain environment, and makes AUV fulfill its mission objectives along the collision-free path. Based on the theory and the application of GA, this paper presents a global path planning method for AUV based on GA and domain knowledge in a large-scale chart. The grid method is adopted to set up a discrete space model for path planning based on known chart data, and each data structure of a grid stores some property such as digital elevation, permit and so on. A kind of decimal grid-coordinate coding scheme which adopting a variable length chromosome encoding is presented. The generating method of initial population, the fitness evaluation function, the evolve strategy and some superiority genetic operators are all designed and introduced in detail. And some measures are also adopted to improve the searching capability and to speed up convergence of the algorithm. The planning results show that the GA path planner has some advantages such as more stronger ability for searching a global optimized and viable path, efficiency superiorly, convergence rapidly, path descried more simply and perspicuous. The GA path planner based on domain knowledge becomes an important element of AUV autonomy ability and can be potentially applied as an on-line path planner for AUV. 2004 IEEE.