NRI: Enhancing Autonomous Underwater Robot Perception for Aquatic Species Management

水下领域几乎占地球的五分之四，对人类福祉、生态平衡以及社会经济繁荣具有重大影响。然而，它对自主系统的探索提出了重大挑战，特别是在感知和定位能力方面。许多应用，例如，在环境监测和生态系统保护方面，使用自主水下航行器（AUV）将大大受益，提高效率，生产力和安全性。特别是，保护水下生态系统对于维持适当的生态平衡至关重要，是一项重要的工作，自主水下系统可以成为宝贵的工具。该项目将整合机器人和海洋保护生物学方面的专业知识，为自主水下机器人感知和导航创造新的能力，使机器人能够识别，跟踪和定位水生物种。该研究团队将利用现场试验、出版物、软件、设计和数据集发布，以及会议中的教程和研讨会，在一个负担得起的开源自主机器人平台上整合机器人视觉、学习和本地化方面的理论进展。该研究团队还将把研究成果纳入课程，以更好地为学生在机器人和智能系统方面的工作做好准备，同时推进保护生物学的知识。调查人员将特别注意从代表性不足的群体中招募学生加入研究小组。该项目将整合机器人和海洋保护生物学方面的专业知识，为自主水下机器人感知和导航创造新的能力，使机器人能够识别，跟踪和定位水生物种。这项研究将使AUV能够在沿岸的栖息地充当有效的机器人助手，以管理这些生态系统。具体而言，研究人员将创建方法来（1）增强多模式水下图像，专门用于水生物种的稳健检测，（2）使用零和/或一次性学习来识别具有有限训练图像的水生物种，以及（3）联合收割机结合声学和测深方法，以实现准确的水下机器人定位。研究成果将在适当的实地地点进行单独评估，并作为一个综合、连贯的系统在水下航行器上进行评估。研究人员补充说， 水下机器人和水生生态学方面的专业知识。该研究团队所拥有的各种专业知识的整合将导致水下自主机器人技术的重大进步，并使AUV能够在具有重要生态和经济意义的领域中使用。该奖项反映了NSF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 underwater domain takes up almost four-fifths of the planet and has significant bearing on human well-being, ecological balance, and accordingly, socio-economic prosperity. However, it poses significant challenges to exploration by autonomous systems, particularly in perception and localization capabilities. Many applications, e.g., in environmental monitoring and ecosystem preservation, would benefit significantly from using autonomous underwater vehicles (AUVs), with improved efficiency, productivity, and safety. In particular, the preservation of the underwater ecosystem is of utmost importance in maintaining proper ecological balance and is a significant undertaking where autonomous underwater systems can be valuable tools. This project will integrate expertise in robotics and marine conservation biology to create novel capabilities for autonomous underwater robotic perception and navigation which will make it possible for robots to identify, track, and localize aquatic species. The research team will integrate theoretical advances in robot vision, learning, and localization on an affordable, open-source autonomous robotic platform using field trials, publications, software, and design and dataset releases, in addition to tutorials and workshops in conferences. The research team will also incorporate research results into course curricula to better prepare students for jobs in robotics and intelligent systems while advancing knowledge in conservation biology. The investigators will pay particular attention to recruit students from underrepresented groups to the research team. This project will integrate expertise in robotics and marine conservation biology to create novel capabilities for autonomous underwater robotic perception and navigation which will make it possible for robots to identify, track, and localize aquatic species. This research will enable AUVs to act as effective robotic assistants in littoral habitats to manage these ecosystems. Specifically, the investigators will create methods to (1) enhance multimodal underwater imagery specifically for robust detection of aquatic species, (2) use zero- and/or one-shot learning for identifying aquatic species with limited training imagery, and (3) combine acoustic and bathymetric methods for accurate underwater robot localization. The research outcomes will be evaluated both individually and as an integrated, coherent system onboard underwater vehicles at appropriate field locations. Investigators have complementary expertise in underwater robotics and aquatic ecology. The integration of the diverse expertise possessed by the research team will lead to major advances in underwater autonomous robotics and enable the use of AUVs in a domain of significant ecological and economic importance.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.