1.Key Laboratory of Transportation Intelligent Operation and Maintenance Technology and Equipment,Ministry of Education,East China Jiaotong University,Nanchang 330013,China
2.Key Laboratory of Vehicle Intelligent Equipment and Control of Nanchang City,East China Jiaotong University,Nanchang 330013,China
3.Pen -Tung Sah Institute of Micro-Nano Science and Technology,Xiamen University,Xiamen 361102,China
To ensure accurate motion control and safe operation of unmanned engineering vehicles in complex terrain, the construction of high-precision semantic map is one of the key tasks. Firstly, the unstructured terrain scanning was performed by UAV tilt photography, and a continuous terrain point cloud model was obtained by implementing the filter denoising and hole completion. Then, a Sobel-G operator was introduced to establish the digital gradient model, and combined with actual vehicle driving state in vehicle-terrain dynamic model, the key safety-assured semantic information was further extracted. Finally, a multi-layer map is constructed, which including terrain point cloud model, digital elevation model, 2D RGB model, digital gradient model, risk obstacle and driving safety semantic informations. Experimental verification is conducted in an unstructured terrain environment, and the effectiveness of semantic map is validated through terrain mapping and semantic segmentation. The experimental results show that the intersection over union of risk obstacle semantic segmentation reaches 81%. The extended steady-state margin angle based on the vehicle-terrain coupling model accurately characterizes the unstable motion state of engineering vehicles and can effectively achieve the semantic segmentation of regional driving safety levels. This work further enriches the semantic information layer of unstructured terrain and provides more decision-making basis for the autonomous driving of unmanned engineering vehicles, which improves the operating efficiency of autonomous construction.
非结构地形环境的高精地图建模方法主要有在线实时扫描与离线测绘建模等[2]。其中,SLAM(Simultaneous localization and mapping)技术是在线实时扫描最常用的地图建模方法,广泛应用于移动机器人的实时路径规划,采用的传感器通常包含激光雷达与视觉相机[3]。为了实现轻量化的SLAM系统,Yi等[4]采用无人机搭载激光雷达进行里程计实时建图任务,有效提高了建图精度及效率。Zhang等[5]使用2D激光雷达构建了室内机器人的SLAM系统,实现了高效的路径规划决策任务。视觉相机建模技术虽然在精度上不及激光雷达所提供的点云模型,但其低能耗和轻便性使其在小型机器人系统中得到成熟应用。Matthies等[6,7]对双目视觉在火星探测车中的应用进行了系统阐述,视觉地形建模为探测车在着陆和巡航过程中提供了准确的路径规划与导航控制信息。此外,双目视觉技术已成为野外机器人实时扫描及建图的主流技术方案之一。Ma等[8]通过融合双目视觉与惯性测量单元数据,成功实现了机器狗在复杂野外环境中的自主导航操作。Sock等[9]使用3D激光雷达和相机在线构建了栅格地图,估计了非结构地形的可通行性,结果表明结合激光雷达和相机能够提供更多互补信息。
JiangK, YangD G, LiuC R, et al. A flexible multi-layer map model designed for lane-level route planning in autonomous vehicles[J]. Engineering, 2019, 5(2): 305-318.
[2]
BaoZ B, HossainS, LangH X, et al. A review of high-definition map creation methods for autonomous driving[J]. Engineering Applications of Artificial Intelligence, 2023, 122: No. 106125.
[3]
ChenW F, ZhouC J, ShangG T, et al. SLAM overview: from single sensor to heterogeneous fusion[J]. Remote Sensing, 2022, 14(23): 6033.
[4]
YiS Q, LyuY, HuaL, et al. Light-LOAM: a lightweight lidar odometry and mapping based on graph-matching[J]. IEEE Robotics and Automation Letters, 2024, 9(4): 3219-3226.
[5]
ZhangX X, LaiJ J, XuD L, et al. 2D Lidar-based SLAM and path planning for indoor rescue using mobile robots[J]. Journal of Advanced Transportation, 2020: No.8867937.
[6]
MatthiesL, MaimoneM, JohnsonA, et al. Computer vision on Mars[J]. International Journal of Computer Vision, 2007, 75(1): 67-92.
[7]
ChengY, MaimoneM W, MatthiesL. Visual odometry on the Mars exploration rovers-a tool to ensure accurate driving and science imaging[J]. IEEE Robotics & Automation Magazine, 2006, 13(2): 54-62.
[8]
MaJ, BajracharyaM, SuscaS, et al. Real-time pose estimation of a dynamic quadruped in GPS-denied environments for 24-hour operation[J]. The International Journal of Robotics Research, 2016, 35(6): 631-653.
[9]
SockJ, KimJ, MinJ, et al. Probabilistic traversability map generation using 3D-LIDAR and camera[C]∥ IEEE International Conference on Robotics and Automation(ICRA), Stockholm, Sweden, 2016: 5631-5637.
LiLi, BaoYu-jian, YangWen-chen, et al. Standardized constructing method of a roadside multi-source sensing dataset[J]. Journal of Jilin University (Engineering and Technology Edition), 2025,55(2):529-536.
[14]
HuangC, ZhangH M, ZhaoJ H. High-efficiency determination of coastline by combination of tidal level and coastal zone DEM from UAV tilt photogrammetry[J]. Remote Sensing, 2020, 12(14): No.2189.
HanYou-mei, AiLin, LiuDong-ming, et al. Research on 1: 500 topographic map measurement technology of urban area based on light and small UAV[J]. Engineering of Surveying and Mapping, 2023, 32(4): 51-58.
[17]
HuX Y, LiD. Research on a single-tree point cloud segmentation method based on UAV tilt photography and deep learning algorithm[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13: 4111-4120.
[18]
ChenW, ZhuQ Y, ZhengX P, et al. A feature fusion-based optimization approach for unstructured terrain modeling in agriculture[J]. Computers and Electronics in Agriculture, 2024, 225: No.109276.
WangShun-jie, LiuQian-jie, LiChen-yang, et al. Lateral dynamic modeling of articulated wheeled loader[J]. Journal of Mechanical & Electrical Engineering, 2021, 38(12): 1529-1537.
[21]
ZhangX Y, LiuQ J, LiuJ, et al. Using gyro stabilizer for active anti-rollover control of articulated wheeled loader vehicles[J]. Journal of Systems and Control Engineering, 2021, 235(2): 237-248.
LiuJun. Research on high precision mapping and positioning for unmanned engineering vehicle[D]. Xiamen: School of Aerospace Engineering, Xiamen University, 2021.
[24]
HuangT C, ChenX W, HuH S, et al. Terrain-adaptive motion planner for articulated construction vehicles in unstructured environments[J]. Automation in Construction, 2024, 168: No.105864.
[25]
GianinettoM. Automatic digital terrain model generation using Cartosat-1 stereo images[J]. Sensor Review, 2008, 28(4): 299-310.
[26]
WangL, YouZ H, LuW, et al. Attention-aware Sobel graph convolutional network for remote sensing image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: No.4409912.
[27]
ZhuQ Y, YiJ, ChenH Y, et al. Lateral stability simulation and analysis for wheel loaders based on the steady-state margin angle[J]. International Journal of Modelling, Identification and Control, 2014, 22(3): 185-194.
[28]
TongC H, GingrasD, LaroseK, et al. The Canadian planetary emulation terrain 3D mapping dataset[J]. International Journal of Robotics Research, 2013, 32(4): 389-395.