Abstract
This paper presents a comprehensive analysis of state-of-the-art software solutions designed for the perception of unstructured outdoor environments, with a focus on their applicability in goal-oriented navigation for off-road unmanned ground vehicles (UGVs). The analysis encompasses the evaluation of various sensors for UGV navigation, comparison of methods for extracting perceptual information from sensor data, and the development of a perception system based on the findings. The proposed system leverages advancements in simultaneous localization and mapping (SLAM), visual odometry, and traversability segmentation using 3D LiDAR and visual systems. The analysis suggests a phased approach for the development of autonomous navigation, where the initial phase relies on complex SLAM and teleoperation assistance, and the subsequent phase introduces an autonomous mode utilizing semantical terrain representation.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
