PFI-RP: Commercialization of Automotive Lane Line Detection Software for Snow-Covered Roads

创新研究伙伴关系（PFI-RP）项目的更广泛影响/商业潜力是在恶劣天气条件下为公众改善运输安全性和可用性。此外，该项目旨在开发和商业化可应用于其他应用的新型计算机视觉和机器学习技术，促进汽车行业参与解决恶劣天气驾驶问题，并使未被充分代表的社区参与创业。实现车辆在恶劣天气下的安全运行，与自动化车辆的广泛部署和特征有进一步的关系，包括提高应对气候变化的效率，改进地面车辆的军事应用，以及更强大的交通网络设计和管理。将招募来自代表性不足社区的学生，并通过与全校现有组织的合作，接受转化研究和创业方面的培训。拟议中的项目旨在开发针对恶劣天气的自动车辆计算机视觉，尽管通常被认为是一种必要的产品，但在市场上受到的关注相对较少。该技术利用汽车摄像头、计算机视觉、系统工程原理和改进的机器学习方法，而不过度依赖深度学习。将计算机视觉算法转化为具有嵌入式硬件的全自动车辆系统和商业化需要多学科的方法，需要学术、工业和企业的能力。四个关键技术障碍包括数据扩展、模块改进、车辆集成和嵌入式硬件的使用。由于确定的最小可行产品是为在雪地中安全自动驾驶而设计的，因此该团队正在追求冬季里程碑，包括大型数据集收集、模块改进以及相关车辆集成演示，以及与潜在业务合作伙伴和客户进行完整的最小可行产品演示。该项目的目标是成功商业化用于积雪路面的汽车车道线检测软件。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 is improved transportation safety and usability for the general public during adverse weather conditions. Additionally, this project seeks to develop and commercialize new computer vision and machine learning techniques that can be applied to other applications, facilitating engagement from the automotive industry in addressing inclement weather driving problems and enabling entrepreneurial engagement from underrepresented communities. Achievement of safe vehicle operation in adverse weather has further implications associated with the widespread deployment of automated vehicles and features including improved efficiency to combat climate change, improvements in ground vehicles for military applications, and a stronger transportation network design and management. Students from underrepresented communities will be recruited and receive training in translational research and entrepreneurship through engagement with existing university-wide organizations. The proposed project seeks to develop automated vehicle computer vision for adverse weather which, despite being commonly identified as a needed product, has received relatively little attention in the marketplace. The proposed technology utilizes automotive cameras, computer vision, systems engineering principles, and refined methods of machine learning without an overreliance on deep learning. The translation and commercialization of computer vision algorithms into a fully automated vehicle systems with embedded hardware requires a multidisciplinary approach requiring academic, industrial, and entrepreneurial competence. The four key technical barriers include data scale-up, module improvements, vehicle integration, and use of embedded hardware. Because the identified minimum viable product is designed for safe automated driving in snow, the team is pursuing winter season milestones that include large dataset collection, module improvements with an associated vehicle integration demonstration, and a complete minimum viable product demonstration with potential business partners and customers. The project goal is successful commercialization of an automotive lane line detection software for snow-covered roads.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.