From Exploring to Optimal Path Planning: Considering Error of Navigation in Multi-Agent Mobile Robot Domain

In this paper a complex path planning model is presented. The model building starts by mapping the totally unknown environment, then the geometrical map of the “work space” is built-up. Afterwards, based on the determined error of navigation, the reduced free space of the environment is created. The path planning process is begins with this reduced environment. At first it determines some cruises (corridors) for the multi-agent traffic, and then comes the builds up the graph-map of the cruises. Based on the weighting of the graphs, the algorithm selects the time-optimal, dynamically optimal, and collisionfree path, from among the possible ones, connecting the starting and docking positions. The final path is created with fitting some B-Spline curves to the selected graph-like one in consideration of movements of the other agents.