System Theoretic Approach to Autonomous Vehicle Dynamic Formation

摘要

A dynamic network of cooperating agents is characterized by a spatially distributed set of dynamic nodes or agents which coordinate to perform the mission objectives. These coordinated dusters could be composed of an array of satellites constructing a large aperture radar or a swarm of UAV's used to suppress enemy air defenses. The mission objectives are to be achieved in the presence of large uncertainties due largely to a hostile environment. Within this context, nod may fail at various levels, measurements may be highly corrupted and communication channels may be severely limited due to jamming. Communication links are further challenged due to power constraints and large spatial dispersion, producing tradeoffs between uncertain information, latency and bandwidth constraints. A decision and allocation process appears computationally intractable, especially if mechanized using a centralized architecture. Over the past three years, important insights have been gained and significant progress has been made on certain basic issues associated with the development of decentralized fault detection and identification, and control algorithms for non-classical information patterns. These efforts allowed an appreciation of the complexity of the decentralized problem, but more importantly they showed the directions that should be taken to make significant progress in the fundamental issues of distributed estimation, analytical redundancy management, and control. In particular, control issues involving the data transmission through noisy channels are explored. Furthermore, the problem of detecting faults in local agents and the development of a decentralized methodology for distributed redundancy management was addressed and some resolution to these problems obtained. These results have given new direction in the development of a theory for the control of dynamic networks.