GA-Based Motion Planning For Underwater Robotic Vehicles

This paper discusses GA-based motion planning in 3D space for underwater robotic vehicles. We propose a genetic algorithm (GA) for path planning which generates a collision-free path in an environment with two kinds of obstacles: Solid and hazardous obstacles such that a path cannot intersect solid obstacles while it can intersect hazardous obstacles at the expense of extra costs. The most important advantage of our GA-based approach is the adaptivity in the sense that a GA can respond to environmental changes (e.g., encounter with an unknown obstacle) and adjust a path \globally" to the new environment. Therefore, the GA approach can incorporate on-line motion planning with o -line motion planning. In addition, the coding used in the GA can be extended so that path planning and trajectory planning are solved simultaneously.