Terrain Classification with Vegetation Augmentation for Outdoor Autonomous Mobile Robot

Terrain classification in outdoor environments is a critical area of research in the field of autonomous navigation. In situations such as forestry, which feature rigid structures, uneven terrain, pits, and cluttered grass, accurate terrain classification enables autonomous mobile robots to understand and navigate these complex terrains effectively. This paper proposes the height-length-density-vegetation (HLDV) terrain classifier, which extends the traditional height-length-density (HLD) classifier by utilizing the same features—height difference <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\boldsymbol{h})$</tex>, length <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\boldsymbol{l})$</tex>, and density <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\boldsymbol{d})$</tex> —and incorporating an additional vegetation <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(v)$</tex> feature to classify safe and dangerous areas for the robots. Experimental results show that the HLDV classifier, particularly when using a power law function for the vegetation feature, significantly improves the robot's ability to navigate through vegetated areas.