Autonomous Agent Navigation Based on Textural Analysis

Navigating an autonomous agent outdoors can be a challenging task. As a basis for this research, the autonomous agent in question is that of an autonomous robotic lawnmower. Mowing a lawn can be a difficult, tedious, and sometimes hazardous chore for the human operator. The goal of this paper is to report on ongoing research of using the visual property of texture to design a system which gives an autonomous lawnmower robot the ability to mow a lawn similar to how a human would (in some type of pattern). Previous autonomous lawnmowers that have been designed, the LawnNibbler and LawnShark for example, rely on the principle of randomness to mow in a confined area. While this method will work, it will take an extraordinary long amount of time. This paper first presents an introduction into what texture is and how it can be used to perform navigation. A system for analyzing textures is then presented. Finally, the current state and future research of this project are then presented.