S&AS: FND: COLLAB: Planning and Control of Heterogeneous Robot Teams for Ocean Monitoring

海洋和沿海环境中的机器人被设想用于商业、科学探索、搜索和救援以及许多其他任务。然而，在这样的环境中，机器人面临着巨大的挑战。车辆必须在只有近场感知的大惯性影响和干扰的环境中作用和影响变化。再加上我们对海洋动力学的有限了解，以及缺乏可访问和高质量的海洋流动数据，这些障碍使得机器人技术在这些不同的应用中的使用变得极其困难。该项目实现了一种综合的、不同种类的机器人方法，用于大规模海洋监测，以缓解环境和搜救行动。它能够对海洋环境中感兴趣的各种物理、化学和/或生物过程进行数据驱动的跟踪和测绘，例如跟踪污染物扩散或失踪的飞机。该项目显著提高了海洋搜索和监测技术的最先进水平，帮助我们了解和利用洋流，并改善世界海洋的健康。该项目的成果被纳入教育，通过私人投资机构的课程，指导学生的研究，并扩大现有的K-12外联关系。该项目创造了关于机器人能够更好地监控、感知和在动态和不确定环境中操作的新方法的基础知识。该项目为不同的监测机器人团队开发了新的方法，以通过当前与环境的交互来改进他们的环境模型；同时收集数据，将其处理并同化到现有模型中，并在该模型上进行计划；接受高级指令，并将环境监测等面向目标的指令转化为适当的计划，以便通过底层系统架构进行感知、推理、通信和行动；以及监测它们的行动，进行自主优化和重新配置。提议的不同种类的机器人团队包括在海空界面提供样本的水面车辆和创建流动模型并在团队中充当中间人的空中机器人。该方法的分层结构利用了由拉格朗日相干结构定义的自然边界来创建分布式感知框架。

Robots in marine and littoral environments are envisioned for commerce, scientific exploration, search and rescue, and many other tasks. However, robots in such environments face significant challenges. Vehicles must act and effect change in environments with large inertial effects and disturbances, with only near-field perception. Coupled with our limited understanding of ocean dynamics and the lack of accessible and high-quality ocean flow data, these obstacles make the use of robotics technology in these varied applications extremely difficult. This project realizes an integrated, heterogeneous robotic approach towards large-scale ocean monitoring for environmental mitigation and search and rescue operations. It enables data-driven tracking and mapping of various physical, chemical, and/or biological processes of interest in marine environments, such as tracking contaminant dispersion or missing aircraft. This project significantly improves the state of the art in ocean search and monitoring technology, helping us understand and harness ocean currents, and improve the health of the world's oceans. Results from the project are integrated into education, through the PIs' courses, mentoring students on research, and expanding an existing K-12 outreach relationship. The project creates fundamental knowledge about new ways that robots can better monitor, sense, and operate in dynamic and uncertain environments. The project develops new methods for heterogeneous teams of monitoring robots to improve their environment model through current interactions with the environment; concurrently collect data, process and assimilate it into the existing model, and plan on that model; accept high-level instruction and translate goal-oriented directives such as environmental monitoring into a suitable plan for sensing, reasoning, communicating, and acting through the underlying system architecture; and monitor their actions, optimize, and reconfigure autonomously. The heterogeneous team of robots proposed includes surface vehicles providing samples at the air-sea interface and aerial robots creating flow models and acting as intermediaries within the team. The hierarchical structure of the approach takes advantage of the natural boundaries defined by Lagrangian coherent structures in the creation of a distributed sensing framework.