Motion planning in isothetic workspaces

This paper explores the issue of algorithmic motion planning in 2D isothetic workspaces. In an isothetic workspace sides of an object are parallel to the principal axes. Motion planning is a complex exercise pertaining to the task-achieving behavior of a robot system. The complexity primarily arises due to the shape of the moving object and its rotation. Complexity-theoretic studies have shown that the problem of optimal motion planning for a non-point moving object with rotation is intractable in a generalized setting. Thus we have chosen to work with a simplified workspace and have obtained algorithmic solutions. The framework developed in this paper is based on an explicit representation of freespace using overlapping rectangles. The freespace can be visualized as a network of linked rectangular regions. The moving object maneuvers through these regions connecting the source and destination points. However, not all the regions and their overlaps are large enough to allow a collision-free motion, consequently restricting some paths for the moving object. We have introduced the notion of characterizing the representation at a higher level in terms of motion actions, i.e. translation and rotation, in graph model. The properties of this graph enable the application of fast graph search algorithms leading to efficient solutions