GRID MODELING OF ROBOT CELLS - A MEMORY-EFFICIENT APPROACH

This article introduces a new approach to modeling robot workspace in two dimensions (and its extension to 2(1/2) dimensions and to mobile objects). The scheme proves to be attractive because it requires less memory space and is execution-efficient. It is very useful for real-time applications. The grid modeling consists in dividing the space into horizontal and vertical strips whose intersection will form elementary cells. A function giving the access conditions (either a binary one allowing access or nonallowed access, or a more elaborate one including other elements (task, geometries)) is associated to each cell. The size of the adaptive grid will depend on the criterion of the obstacle minimal approach by the robot. Then a method for path planning is developed. The Lee and A(*) algorithms being the starting point, the method has been conceived by executing mask operations. So the method requires less memory space and allows an admissible trajectory to be quickly found.