Distributed Illumination Problems for Visually-Guided Agents

拟议的研究重点是解决照明问题的算法，这是一类几何优化问题，重点是在哪里放置，或如何移动，以照亮感兴趣的环境或目标的方式的警卫组。我们的目标是解决这类几何优化问题，通过视觉引导代理，即，该技术方法基于一套新的“增量分区和部署”算法，有效地解决了具有可见性传感器的机器人警卫的经典艺术画廊问题。 我们的解决方案依赖于反映环境拓扑结构的新颖几何结构，探索这些结构的导航算法，以及通过本地传感和通信收集的信息的分布式算法。 广泛的研究目标是为各种照明和运动协调问题设计分布式算法。通过考虑环境的层次结构、各种照明模型和不同的任务，产生了有用的场景。 重点是在城市环境中，带有摄像头和其他可控传感器的机器人组将提供无处不在的覆盖和传感。 示例应用可以包括民事救援行动和自适应空中监视。 这项技术有望通过促进国家安全，工业自动化和新颖的商业应用中的许多关键任务来影响整个社会。 这项研究将通过定期出版物以及通过开发合作移动的机器人技术的讲义和课程来实现广泛传播。 PI将开发用于分布式照明问题、几何优化和算法可视化的软件。PI将监督本科生团队执行与本提案的广泛主题相关的机器人和控制高级设计项目，包括传感器网络，移动机器人控制和视觉传感。 PI计划尽可能广泛地宣传与提案相关的项目，以吸引代表性不足和少数民族学生。

The proposed research focuses on algorithms to solve illumination problems,which are a class of geometric optimization problems that focus on where toplace, or how to move, groups of guards in such a way as to illuminateenvironments or targets of interest. The objective is to solve this classof geometric optimization problems, through visually-guided agents, i.e.,robotic agents with line-of-sight sensors and line-of-sight communication.The technical approach is based on a novel set of "incremental partitionand deployment" algorithms that effectively solve the classic art galleryproblem for robotic guards with visibility sensors. Our solutions rely onnovel geometric structures that reflect the environment topology, onnavigation algorithms to explore these structures, and on distributedalgorithms to manage the information gathered via local sensing andcommunication. The broad research objective is to design distributedalgorithms for various illumination and motion coordinationproblems. Useful scenarios arise by considering a hierarchy ofenvironments, a variety of illumination models, and different tasks.The proposed research will impact the emerging technologies of robotic andsensor networks. The focus is on situations where groups of robots withcameras and other controllable sensors will provide pervasive coverage andsensing in urban environments. Example applications could include civilrescue operations and adaptive aerial surveillance. This technology holdsthe promise to affect society at large by facilitating a number of criticaltasks in national security, industrial automation, and novel commercialapplications. This research will achieve broad dissemination throughregular publications as well as through the development of lecture notesand a course on cooperative mobile robotics. The PI will develop softwarefor distributed illumination problems, geometric optimization and algorithmvisualization. The PI will supervise teams of undergraduate studentsperforming senior design projects on robotics and controls relatedto the broad themes of this proposal, including sensor networks, mobilerobot control, and visual sensing. The PI plans to advertise opportunitiesto work on proposal-related projects as widely as possible to attractunder-represented and minority students.