A comprehensive survey of unmanned ground vehicle terrain traversability for unstructured environments and sensor technology insights
Semih Beyçimen, Dmitry Ignatyev, Argyrios Zolotas
Abstract
This article provides a detailed analysis of the assessment of unmanned ground vehicle terrain traversability. The analysis is categorized into terrain classification, terrain mapping, and cost-based traversability, with subcategories of appearance-based, geometry-based, and mixed-based methods. The article also explores the use of machine learning (ML), deep learning (DL) and reinforcement learning (RL) and other based end-to-end methods as crucial components for advanced terrain traversability analysis. The investigation indicates that a mixed approach, incorporating both exteroceptive and proprioceptive sensors, is more effective, optimized, and reliable for traversability analysis. Additionally, the article discusses the vehicle platforms and sensor technologies used in traversability analysis, making it a valuable resource for researchers in the field. Overall, this paper contributes significantly to the current understanding of traversability analysis in unstructured environments and provides insights for future sensor-based research on advanced traversability analysis.