无人机看，无人车做：航空影像与虚拟示教实现地面车辆零样本复现
UAV See, UGV Do: Aerial Imagery and Virtual Teach Enabling Zero-Shot Ground Vehicle Repeat
- University of Toronto Robotics Institute（多伦多大学机器人研究所）
- University of Toronto Institute of Aerospace Studies（多伦多大学航空航天研究学院）
- Ingenuity Labs（Ingenuity实验室）
- Department of Electrical and Computer Engineering（电气与计算机工程系）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本研究提出VirT&R框架，通过航空影像训练NeRF构建虚拟环境定义路径，结合现有LT&R框架实现无GPS环境下UGV零样本自主导航，性能接近传统方法且无需实地人工示教。
AI中文摘要：
本文提出虚拟示教与复现（Virtual Teach and Repeat，VirT&R）：这是示教与复现（Teach and Repeat，T&R）框架的扩展，可在无GPS、未遍历的环境中实现地面车辆的零样本自主导航。VirT&R利用针对目标环境采集的航空影像训练神经辐射场（Neural Radiance Field，NeRF）模型，从而提取密集点云和带照片纹理的网格。NeRF网格被用于构建环境的高保真仿真环境，以操控无人地面车辆（unmanned ground vehicle，UGV）在虚拟环境中定义期望路径。随后，通过使用沿路径关联的NeRF生成点云子图以及现有的激光雷达示教与复现（LiDAR Teach and Repeat，LT&R）框架，即可在实际目标环境中执行任务。我们在超过12公里的自动驾驶数据上对VirT&R的可复现性进行了基准测试，使用物理标记获取仿真到现实的横向路径跟踪误差，并与LT&R进行对比。VirT&R在两种不同环境中测得的均方根误差（RMSE）分别为19.5厘米和18.4厘米，略小于测试所用机器人的一个轮胎宽度（24厘米），对应的最大误差分别为39.4厘米和47.6厘米。该结果仅使用NeRF生成的示教地图就得以实现，证明VirT&R具备与LT&R相近的闭环路径跟踪性能，但无需人类在实际环境中手动向UGV示教路径。
英文摘要：
This paper presents Virtual Teach and Repeat (VirT&R): an extension of the Teach and Repeat (T&R) framework that enables GPS-denied, zero-shot autonomous ground vehicle navigation in untraversed environments. VirT&R leverages aerial imagery captured for a target environment to train a Neural Radiance Field (NeRF) model so that dense point clouds and photo-textured meshes can be extracted. The NeRF mesh is used to create a high-fidelity simulation of the environment for piloting an unmanned ground vehicle (UGV) to virtually define a desired path. The mission can then be executed in the actual target environment by using NeRF-generated point cloud submaps associated along the path and an existing LiDAR Teach and Repeat (LT&R) framework. We benchmark the repeatability of VirT&R on over 12 km of autonomous driving data using physical markings that allow a sim-to-real lateral path-tracking error to be obtained and compared with LT&R. VirT&R achieved measured root mean squared errors (RMSE) of 19.5 cm and 18.4 cm in two different environments, which are slightly less than one tire width (24 cm) on the robot used for testing, and respective maximum errors were 39.4 cm and 47.6 cm. This was done using only the NeRF-derived teach map, demonstrating that VirT&R has similar closed-loop path-tracking performance to LT&R but does not require a human to manually teach the path to the UGV in the actual environment.
