Abstract:
Path planning for a robot is to find a set of motions that will lead the robot end-effecter to move between the two positions without colliding with the workspace boundary including singularity manifolds and obstacles.
The standard way to approach the problem is to express it in another space C-space, having independent parameters as coordinates and robot as a single point. The Path planning problem consists of finding a continuous curve representing a path for a single geometric point.
§ This method proceeds in two phases:
§ Domain Mapping Phase
§ Query Phase
§ In Domain Mapping Phase, the non-convex region is mapped onto a convex region. The target convex region is a sphere. Obstacles in the non-convex region are shrunk to a point.
§ In the Query Phase, the path is found out, for given any source and destination point of the robot, inside the region. For this purpose, corresponding positions of the two points in the mapped region are found out. In the mapped region these two points are connected by a straight line. Now inverting the entire path back to the original region, we can get the actual path.
You can look at My Poster on my thesis.
DOMMAIN MAPPING OF CAT HEAD: SHOWING THE PATH ON THE PALM:
