SBIR Phase I: Robust autonomy for robotics-based data collection in surface water environments

这个小企业创新研究（SBIR）第一阶段项目的更广泛的影响/商业潜力在于使用无人水面车辆（USV）在地表水环境中收集信息的效率提高和成本降低。USV有可能增加水体的可访问性，提高覆盖率，并通过将人类从任务中移除而使数据收集更安全。然而，为了充分实现USV收集信息的潜力，需要具有最低限度的人工监督和干预的自主性。这项技术的发展将为普遍使用USV进行环境数据收集打开大门，并最终通过以下方式造福社会：i）更多的信息，这些信息是资源管理和环境政策决策的基础; ii）通过降低成本提高数据收集活动的频率;和iii）通过使人类工作人员远离恶劣环境来提高安全性。第一阶段项目将为地表沃茨环境数据收集的变革方法奠定基础。该项目的智力价值在于解决技术挑战，使无人水面车辆（USV）能够在可能包含各种障碍物的复杂表面沃茨中有效地自主导航，否则这些障碍物将无法进入，或者充其量需要高度的人类监督。这将通过将计算机视觉和机器学习的最新技术推向智能障碍物表征来实现。所提出的框架解释图像或视频的障碍，提取突出的特征，如灵活性或刚性，通过创新使用机器学习，降维，并借用工具的集体行为的分析。通过拟议的框架提供的额外信息层代表了计算机视觉的重大进步，通过赋予USV新的无人驾驶能力，为环境监测和数据收集提供了直接利益。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 broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project lies in the increase in efficiency and reduction in cost of gathering information in surface water environments using unmanned surface vehicles (USVs). USVs have the potential to increase accessibility to bodies of water, enhance coverage, and make data collection safer by removing humans from the task. However, to fully realize the potential of USVs for gathering information, a level of autonomy with minimal need for human supervision and intervention is required. The development of this technology will open the door to the pervasive use of USVs for environmental data collection, and ultimately benefit society through: i) more information upon which to base decisions about resource management and environmental policy; ii) higher frequency of data collection activities through reduced cost; and iii) increased safety by keeping human personnel out of harsh environments.This Small Business Innovation Research (SBIR) Phase I project will lay the foundation for a transformative approach to environmental data collection in surface waters. The intellectual merit of this project lies in addressing technical challenges that will enable unmanned surface vehicles (USVs) to effectively navigate autonomously in complex surface waters that may contain a range of diverse obstacles, which would otherwise be inaccessible or at best require a high degree of human supervision. This will be achieved by advancing the state-of-the-art in computer vision and machine learning toward intelligent obstacle characterization. The proposed framework interprets images or video of obstacles to extract salient features, such as flexibility or rigidity, through the innovative use of machine learning, dimensionality reduction, and tools borrowed from the analysis of collective behavior. This extra layer of information provided through the proposed framework represents a significant advancement in computer vision, which offers direct benefits to environmental monitoring and data collection by empowering USVs with new unmanned capabilities.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.