Robot path planning based on concept lattice

Abstract

Path planning is a popular topic in research on mobile robots. In this paper, we apply the concept lattice to the path planning problem for the first time in the literature to propose a static path planning algorithm. The given grid map is first transformed into the formal context of the grid, and locational relations between rectangular regions are mapped into partial-order relations in the graph of rectangular regions based on a lattice of region concepts. Following this, the path planning problem on the original grid map is transformed into a path searching problem in the graph of rectangular regions. The graph derived from the sublattice of region concepts is used to avoid the high time complexity of applying the complete concept lattice. Finally, a novel path planning algorithm that can generate a path consisting of rectangular regions is constructed. This path can also be converted into the path consisting of inflection points. The results of simulations verified the effectiveness of the proposed method. It can generate a path with significantly fewer inflection points than the A* algorithm for a four-direction (left, right, front, back) mobile robot.