Autonomous Robotic Systems

点击翻译按钮获取中文摘要

The utilization of robots has been shifting from their principal base of manufacturing toward household, healthcare, security, and exploration applications. Future challenges in the field will, thus, include the development of autonomous robotic systems that can interact with their environment on different levels. Hence, this grant application is in support of our research on the development, planning, and control of autonomous multi-agent systems with a contemporary focus. We expect to make notable advancements on two synergetic sub-themes: (i) development of multi-agent sensing methodologies for the localization, tracking, and recog­ni­tion of time-varying-geometry objects, and (ii) development of multi-agent team-deployment and motion-coordination methodologies for the tracking, interception, and manipulation of these dynamic objects (i.e., targets). Sub-Theme 1:   Our goal is to build on our ongoing work, on multi-camera active-vision systems, for the development of generic on-line sens­ing methodologies for the localization, tracking and recognition of deformable objects in cluttered, dynamic environments. As the first objective for this Sub-Theme, our focus will be on developing novel techniques for the shape recovery of unknown markerless deformable objects through fusion of visual hulls and range data, integrated with state prediction. The primary motivation in markerless shape recovery is removing potential dependency on a database of a priori known object models. As the second objective, we will develop robust optimization processes for multi-camera reconfiguration – an objective to be addressed within the framework of Sub-Theme 2.     Sub-Theme 2:   Our goal is to build on our ongoing work, on the on the development of multi-agent team-deployment and motion-coordination strategies, for the tracking, interception, and manipulation of dynamic targets. As the first objective for this Sub-Theme, our focus will be on developing novel deterministic centralized methods for the multi-agent deployment and coordination problem for trackable targets, whose location can be viewed and their motion predicted on-line – an objective interrelated to Sub-Theme 1. As the second objective, our focus will be on developing novel deterministic methods for the multi-agent deployment and coordination problem for intercepting untrackable targets – e.g., robotic search and rescue in the wilderness. For balanced task allocation, we will first investigate decomposition approaches, where a global specification is decomposed into sub-specifications for individual agents, and corresponding local controllers are developed. As the third objective, our focus will be on the implementation of high-dexterity parallel mechanisms with reconfigurable topologies that allow accurate manipulation of targets in relatively limited workspaces – e.g., MRI-guided robotic surgery.