PFI-RP: Automatic detection of buried explosives using a multi-sensor smart aerial vehicle

该创新研究伙伴关系（PFI-RP）项目具有更广泛的影响/商业潜力，旨在通过使用传感器融合的智能无人飞机自动探测埋藏爆炸物，协助人道主义、科学和国防事业。在受雷区影响最严重的国家，不一致的埋地炸药探测技术使得修复工作进展缓慢，并阻碍了农业发展。该技术通过在单个平台上使用多个传感器，减少了进行自动空中测量所需的操作人员数量，从而降低了埋地爆炸物探测的成本。该项目的教育影响包括项目博士生的创业培训，以及通过项目技术原型的实际演示，激发对地球物理学、机器学习和工程的兴趣，向代表性不足的社区进行当地科学推广。该项目将通过提供一种快速评估哪套仪器将产生最高的埋藏目标识别精度的方法，进一步加深对近地表（0-2米）的地球物理理解。无人驾驶飞机将使用机器学习产生野外区域的复合反演。通过机器学习将多个地球物理仪器的数据最佳地组合在一起，可以增加推断的置信度。从这个项目中获得的见解可能有助于改进基于地球科学中多个互补的地球物理数据集的推断。在该项目的研发部分使用多种机器学习算法将有助于为未来的应用提供信息。该项目将有助于弥补地球物理数据采集和反演在实时和环境适应性方面的差距。除了地埋爆炸物探测之外，该项目的技术平台和深度学习包将使最终用户能够利用地球物理数据更好地区分地下存在的地质和人造材料，特别是在地面实地工作成本高或危险的环境中。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。

The broader impact/commercial potential of this Partnerships for Innovation - Research Partnerships (PFI-RP) project aims to assist in humanitarian, scientific, and defense endeavors by automatically detecting buried explosives using a sensor-fused smart unmanned aircraft. Inconsistent buried explosives detection techniques have made remediation slow and stymied agricultural development in countries most impacted by minefields. The proposed technology offers an opportunity to decrease the cost of buried explosives detection by employing multiple sensors on a single platform and reducing the number of operators needed to conduct the automated aerial survey. Educational impacts of this project include the entrepreneurial training of the project’s PhD student, and local scientific outreach to underrepresented communities through hands-on demonstrations of the project’s technology prototype to inspire interest in geophysics, machine learning, and engineering.This project will further the geophysical understanding of the near-surface (0-2 meters) by providing a method for quickly evaluating which set of instruments will yield the highest buried target identification accuracy. The unmanned aircraft will produce a composite inversion of a field area using machine learning. Optimally combining data from several geophysical instruments through machine learning can increase the confidence of inferences. Insights from this project may help improve inferences based on multiple, complementary geophysical datasets across the geosciences. Using multiple machine learning algorithms in the research and development portion of this project will help inform future applications. This project will help bridge the gap of geophysical data collection and inversions in a real-time and environment-adaptable way. Beyond the scope of buried explosives detection, this project’s technology platform and deep learning package will enable end-users to better differentiate what geologic and man-made materials are present in the subsurface using geophysical data, especially in environments where ground-based fieldwork is costly or hazardous.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.