CRII: CPS: Modeling Subsurface Features and Connected Autonomous Vehicles as Cyber-Physical Systems for Reciprocal Mapping and Localization

在现代城市的地下存在着迷宫般的地下基础设施，包括水管、公用电缆和天然气管道。该项目建议研究使用配备传感器的自动驾驶汽车来精确绘制这一复杂网络，并探索如何利用地下基础设施作为不变地标来提供可靠和安全的车辆导航。如果成功，这项研究将产生自动化工具，可以更好地绘制城市地下地图，以改善地下基础设施，并在需要挖掘时防止事故发生；以及创造一种新的方法，在自然或人为灾害期间和之后，在混乱和破败的城市地区导航自动驾驶汽车。该项目还将为参与的学生提供职业发展机会，包括来自代表性不足群体的学生。为了实现这些目标，多个联网自动驾驶车辆将配备探地雷达，以收集地下基础设施沿协调轨迹的雷达图。将创建一种新的算法来检测和识别雷达图中的特征，这些特征对应于地下基础设施的元素。不需要的特征，如信号干扰和特征遮挡，将变成有用的线索，聚合多个雷达图，以产生完整的地下地图。基于这些地图，将设计一种新的基于雷达图的里程计方法，根据部分观察到的车辆特征网络来估计联网车辆的全局姿态。将对这些方法进行测试和评估，以确定车辆速度和网络拓扑等关键参数如何影响地图、定位和导航的准确性。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Underneath the ground of modern cities exists a labyrinth of subsurface infrastructure consisting of water pipes, utility cables, and gas lines. This project proposes to study the use of sensor-equipped autonomous vehicles to accurately map this complicated network and explore how to leverage the subsurface infrastructure as invariant landmarks to provide reliable and secure vehicle navigation. If successful, this research will result in automated tools that make better maps of urban subsurface to improve buried infrastructure and prevent accidents when digging is required; as well as create a new means to navigate autonomous vehicle in cluttered and distressed urban areas during and after natural or man-made disasters. This project will also offer career development opportunities for participating students including those from underrepresented groups.To achieve these objectives, multiple connected and autonomous vehicles will be equipped with ground penetrating radar to collect radargrams of subsurface infrastructure along coordinated trajectories. A novel algorithm will be created to detect and identify the signatures in radargram that correspond to elements of the subsurface infrastructure. The unwanted features such as signal interferences and signature occlusions will be turned into useful clues to aggregate multiple radargrams to produce complete subsurface maps. From these maps, a new radargram-based odometry method will be devised to estimate the global poses of connected vehicles based on partially observed vehicle-feature networks. The methods will be tested and evaluated to determine how key parameters such as vehicle speed and network topology affect the accuracy of mapping, localization, and navigation.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.