Fuzzy Traversability Index: A new concept for terrain-based navigation

Fuzzy Traversability Index is introduced in this paper as a new and simple measure for quantifying the ease of traversal of natural terrains by field mobile robots. This index provides a simple means for incorporating the terrain quality data into the robot navigation strategy and is used for terrain-based navigation of held mobile robots. The Traversability Index is expressed by fuzzy sets that quantify the suitability of the terrain for traversal based on its geometrical and physical properties, such as slope, roughness, and hardness. This descriptive representation of terrain traversability in a natural language using linguistic variables and conditional statements is easily comprehensible, more appealing, and closer to human intuition than a mathematical formulation of traversability. A set of fuzzy navigation rules is developed using the Fuzzy Traversability Index to guide the robot toward the safest and the most traversable terrain. In addition, another set of fuzzy rules is developed to drive the robot from its initial position to a user-specified goal position. These two rule sets are integrated in a two-stage procedure for autonomous robot navigation without a priori map-based knowledge about the environment. In the first stage, the traverse-terrain and seek-goal rule sets make their individual, independent recommendations for robot turn and move commands. In the second stage, these recommendations are integrated by using appropriate weighting factors to generate the combined, coordinated recommendation for the robot navigation based on the robot status. Three simulation studies are presented to demonstrate the capability of the mobile robot to reach the goal safely while avoiding impassable terrains. (C) 2000 John Wiley & Sons, Inc.