Coordinated autonomous systems: optimized motion in dynamic and uncertain environments

近年来，在便携式计算、传感和驱动方面取得了令人难以置信的进步。我们现在有能力创造出地面、空中和水下机器人，它们可以在复杂的环境中移动，获取丰富的传感器数据，并在船上处理数据以进行推理、导航和行动。这类机器人的应用正在几个领域出现。在环境监测方面，水下机器人正在增加我们对海洋现象的了解；在物资处理方面，大批机器人正在实现仓库的自动化；在军事方面，无人驾驶飞行器正在改变执行侦察任务的方式。 尽管取得了这些进步，但大多数应用程序的自主性水平仍然很低。在许多情况下，人工操作员执行诸如远程操作之类的低级别任务。在自动化仓库等领域，机器人被要求在高度结构化的环境中操作。之所以使用这些方法，是因为我们目前还没有在动态和不确定的环境中规划可靠和有效的运动的算法。 这项研究计划将开发控制算法，使机器人能够1)基于不完整的环境信息规划运动，以及2)在获得新信息时调整运动规划。我们将专注于证明，即使在存在环境不确定性的情况下，算法也能产生满足预期规范的机器人行为。这项研究将针对三个主要应用领域：1)环境监测，即一组机器人必须收集有关动态环境的信息；2)紧急响应，机器人必须发现并响应紧急事件；以及3)自动化材料处理，机器人必须在复杂的环境中运输物品。

Recent years have witnessed incredible advances in portable computation, sensing, and actuation. We are now capable of creating ground, aerial, and underwater robots that can move through complex environments, acquire rich sensor data, and process the data on-board to reason, navigate, and act. Applications of such robots are occurring in several domains. In environmental monitoring, underwater robots are increasing our understanding of oceanographic phenomena; in material handling, large fleets of robots are automating warehouses; and, in the military, uninhabited aerial vehicles are transforming the way in which reconnaissance missions are performed. Despite these advances, the level of autonomy in most applications is quite low. In many instances, human operators perform low-level tasks, such as remote operation. In areas such as automated warehouses, robots are required to operate in highly structured environments. These approaches are used because we do not currently have algorithms for planning reliable and efficient motion in dynamic and uncertain environments. This research program will develop control algorithms that enable robots to 1) plan motion based on incomplete environmental information, and 2) adapt motion plans as new information becomes available. We will focus on certifying that the algorithms produce robot behaviours that satisfy desired specifications, even in the presence of environmental uncertainty. The research will target three main applications areas: 1) environmental monitoring, where groups of robots must collect information about dynamic environments; 2) emergency response, where robots must discover and respond to urgent events; and 3) automated material handling, where robots must transport items in complex environments.