This project will focus on the specific instance of machine learning demonstrated by LAGR to define and navigate surroundings. It will especially examine the methods used to “see” obstacles and avoid them in a given environment. I will attempt to code an algorithm that adapts its path given obstacles randomly placed in two dimensions.
LAGR uses images taken with the vision system installed on the front of the robot. It has the ability to determine if an area is traversable and places where there are likely paths. One of the issues faced by the designers is the lack of ability to determine the placement of ditches, water, and fences. As it is, LAGR has the ability to “see” for eight meters ahead, and using that information, it plots a best path to take in the direction of its destination.