Traversability estimation based on vehicle characteristics and environmental geometry

In unstructured environments, ensuring the safe operation of autonomous vehicles necessitates precise perception and thorough analysis of the surrounding terrain. Existing traversability analysis methods frequently rely on data from a single sensor and fail to integrate vehicle characteristics with environmental geometric features. This limitation results in weak environmental perception and poor performance in traversability analysis. To address these challenges, this paper proposes an innovative approach. By fusing data from multiple LiDAR sensors, the method constructs a locally dense traversability map, significantly enhancing the completeness and reliability of terrain information.Moreover, this approach combines vehicle characteristics with multiple environmental geometric factors to generate a traversability layer. This integration leads to a more holistic understanding of the terrain, improving the vehicle's ability to navigate complex environments safely. The proposed method demonstrates outstanding performance in traversability estimation, enabling a more accurate evaluation of traversability. This, in turn, provides a solid guarantee for the safe driving of autonomous vehicles in various challenging scenarios.The system underwent rigorous testing in unstructured environments to validate its effectiveness. The test results indicate that the elevation maps and traversability layer generated by this method exhibit significant improvements in both detail and accuracy. These advancements ensure that autonomous vehicles can better perceive and navigate their surroundings, ultimately contributing to safer and more reliable autonomous driving in unstructured and complex terrains.