Hybrid terrain traversability analysis in off-road environments

Date

2022-03-22

Supervisor/s

Journal Title

Journal ISSN

Volume Title

Publisher

IEEE

Department

Type

Conference paper

ISSN

2767-7737

Format

Free to read from

Citation

Leung THY, Ignatyev D, Zolotas A. (2022) Hybrid terrain traversability analysis in off-road environments. In: 2022 8th International Conference on Automation, Robotics and Applications (ICARA), 18 February - 20 March 2022, Prague, Czech Republic

Abstract

There is a significant growth in autonomy level in off-road ground vehicles. However, unknown off-road environments are often challenging due to their unstructured and rough nature. To find a path that the robot can move smoothly to its destination, it needs to analyse the surrounding terrain. In this paper, we present a hybrid terrain traversability analysis framework. Semantic segmentation is implemented to understand different types of the terrain surrounding the robot; meanwhile geometrical properties of the terrain are assessed with the aid of a probabilistic terrain estimation. The framework represents the traversability analysis on a robot-centric cost map, which is available to the path planners. We evaluated the proposed framework with synchronised sensor data captured while driving the robot in real off-road environments. This thorough terrain traversability analysis will be crucial for autonomous navigation systems in off-road environments.

Description

Software Description

Software Language

Github

Keywords

terrain traversability analysis, off-road environments, elevation mapping, semantic segmentation, cost map

DOI

Rights

Attribution-NonCommercial 4.0 International

Relationships

Relationships

Supplements

Funder/s