RII Track-4: Data-Driven Navigation, Path Planning, and Coordination of Mobile Robots in Fluids

RII Track-4：数据驱动的导航、路径规划和流体中移动机器人的协调

• 批准号: 2032522
• 负责人: Zhuoyuan Song
• 金额: $ 18.53万
• 依托单位: University of Hawaii
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-02-01 至 2023-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2032522&HistoricalAwards=false
• 关键词: RII; Track; Data; Driven; Navigation

Autonomous mobile robots, such as unmanned aerial and underwater vehicles, are becoming an essential element in an increasing number of applications that serve the national interest. These intelligent platforms can provide critical in-situ measurement data for weather forecasting and persistent monitoring of the forming and developing of large-scale dynamic events. The capabilities to navigate and sense dynamic fluid environments are fundamental to miniature mobile robots in strong geophysical flows such as hurricanes or ocean currents. These capabilities require the robots to sense, explore, and understand background flow dynamics that are often not amenable to empirical models based on first principles, challenging the existing approaches for robot autonomy. This project aims at enhancing the navigation, path planning, and coordination of mobile robots in dynamic fluid environments by using data-driven system dynamics modeling, estimation, and control. The goals of this fellowship project are to straighten the PI’s robotics research program with intensive training on data-driven dynamics modeling and control techniques at the University of Washington, and bring long-term sustained improvements to the research and education capacity of the University of Hawai‘i system on data science and robotics. This project advances robot navigation, path planning, and coordination in fluid environments, which are fundamental for global ocean sensing and weather forecasting. The proposed research contributes to the foundation of robot autonomy by combing physics-informed, data-driven modeling with classical control and estimation based on first principles. The research objectives are to (1) build the theoretical foundation for fluid-based simultaneous localization and mapping using probabilistic inference, dynamic compressed sensing, and sparse identification of nonlinear dynamics, (2) uncover the connection between finite-horizon optimal trajectories in unsteady flow fields and the underlying coherent flow structures using model predictive control, and (3) identify the optimal swarming laws and emergent swarm dynamics in unsteady fluids using data-driven dynamics learning. The training and synergistic objectives are to establish a mutual student co-advising relationship, develop a new undergraduate course on “Data Science for Engineers” for the University of Hawai‘i, and initiate new collaborations with the eScience Institute and the Applied Physics Laboratory at the University of Washington. The expected outcomes include but are not limited to long-lasting collaborations between the PI and the host, joint journal publications, lightboard lecture videos series for the new course, and strategic plans for co-developing an open-access marine robotics testbed in Hawai‘i. The project impact will be sustained through joint publications, collaborative proposal development, student co-advising, and collaborations between the University of Hawai‘i and the University of Washington on data science research and education.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.

自主移动机器人，如无人驾驶的空中和水下机器人，正在成为越来越多服务于国家利益的应用的基本要素。这些智能平台可以为天气预报和持续监测大范围动态事件的形成和发展提供关键的现场测量数据。导航和感知动态流体环境的能力是微型移动机器人在飓风或洋流等强烈地球物理流动中的基础。这些能力要求机器人感知、探索和理解背景流动态，而这些背景流动态往往不符合基于第一性原理的经验模型，对现有的机器人自主方法提出了挑战。该项目旨在通过使用数据驱动的系统动力学建模、估计和控制来增强移动机器人在动态流体环境中的导航、路径规划和协调。该奖学金项目的目标是通过在华盛顿大学进行数据驱动动力学建模和控制技术的强化培训，理顺PI的机器人研究计划，并为夏威夷大学数据科学和机器人系统的研究和教育能力带来长期持续的改善。该项目推进了机器人导航、路径规划和在流动环境中的协调，这些都是全球海洋传感和天气预报的基础。该研究将物理信息、数据驱动的建模与经典的基于第一性原理的控制和估计相结合，为机器人自主性的研究奠定了基础。研究目标是：(1)利用概率推理、动态压缩传感和非线性动力学稀疏辨识，为基于流体的同时定位和测绘奠定理论基础；(2)利用模型预测控制，揭示非定常流场中有限水平最优轨迹与基本相干流动结构之间的联系；(3)利用数据驱动动力学学习，识别非定常流体中的最优蜂群规律和涌现的蜂群动力学。培训和协同的目标是建立相互的学生共同咨询关系，为夏威夷大学开发一门新的本科课程，即“面向工程师的数据科学”，并启动与电子科学研究所和华盛顿大学应用物理实验室的新合作。预期的结果包括但不限于国际和平协会和东道主之间的长期合作、联合期刊出版物、新课程的灯板讲座视频系列，以及共同开发夏威夷开放访问海洋机器人试验台的战略计划。该项目的影响将通过联合出版物、协作提案开发、学生共同建议以及夏威夷大学和华盛顿大学在数据科学研究和教育方面的合作来持续。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Path Planning for Optimal Coverage of Areas with Nonuniform Importance

• DOI: 10.2514/6.2022-2546
• 发表时间: 2021-10
• 期刊: AIAA SCITECH 2022 Forum
• 作者: Gregory F. Snyder;Sachin Shriwastav;Dylan Morrison-Fogel;Zhuoyuan Song