The introduction of autonomous flight, whether military, commercial, or civilian, into the National Airspace System (NAS) will present significant challenges. Minimizing the impact and preventing the changes from becoming disruptive, rather than an enhancing technology will not be without difficulty. From obstacle detection and avoidance to real-time verification and validation of system behavior, there are significant problems which must be solved prior to the general acceptance of autonomous systems. This paper examines some of the key challenges and the multi-disciplinary collaboration which must occur for autonomous systems to be accepted as equal partners in the NAS. N fiction, autonomous systems do not have a good reputation. From the Jewish folk tale, Golem of Prague , through Mary Shelley's Frankenstein , to Fred Saberhaen's Berserkers series, as well as the infamous Cylons of Battlestar Galactica , automatons are generally not presented as society's benefactors. Even when not terrorizing the world, they are often seen as paranoid − HAL of 2001 and Marvin of The Hitchhikers Guide to the Galaxy − or worse, as bumbling tin men looking for a brain. These literature plots reflect their author's perspective of potential problems that can, or might result from the introduction of autonomous systems into the real world. Although the introduction of autonomous flight into the National Airspace will be technically challenging, the most difficult part may be getting good press for the automatons. To accomplish the smooth introduction of autonomous aircraft into the National Airspace System (NAS) we need significantly more knowledge than simply knowing how to design, implement, test, and verify an autonomous system. We need to understand not only the individual autonomous system, but also how collections of tens or hundreds of them interact with each other, as well as with piloted traffic, all vying for maneuvering room within the NAS. Methods need to be developed to understand how unexpected interactions can occur within the control system of an autonomous aerial vehicle, and the potential for unexpected emergent behavior in the swarms of autonomous unmanned aerial systems (UAS) moving within the NAS. And when the understanding of how to build well- behaved autonomous systems is learned, it will be necessary to develop the capabilities to be able to formally verify and validate that these systems are safe: safe for themselves, safe to other aerial vehicles, and safe to anyone or thing on the ground. But the most daunting problem will be developing the skills and knowledge to make autonomous systems trustworthy. For at the bottom line, allowing autonomous aerial systems into the NAS is all about trusting our systems.

In the last decade, the realization of small, inexpensive, and powerful devices with sensors, computers, and wireless communication has promised the development of massive sized sensor networks with dense deployments over large areas capable of high fidelity situational assessments. However, most management models have been based on centralized control and research has concentrated on methods for passing data from sensor devices to the central controller. Most implementations have been small but, as it is not scalable, this methodology is insufficient for massive deployments. Here, a specific application of a large sensor network for adaptive noise reduction demonstrates a new paradigm where communities of sensor/computer devices assess local conditions and make local decisions from which emerges a global behaviour. This approach obviates many of the problems of centralized control as it is not prone to single point of failure and is more scalable, efficient, robust, and fault tolerant.

Situation management is a rapidly developing science combining new techniques for data collection with advanced methods of data fusion to facilitate the process leading to correct decisions prescribing action. Current research focuses on reducing increasing amounts of diverse data to knowledge used by decision makers and on reducing time between observations, decisions and actions. We propose a new architecture modeled after biological ecosystems where motes are autonomous and intelligent, yet cooperate with local neighborhoods. While situation management research is currently dominated by military applications, advances envisioned for industrial and business applications have similar requirements. NASA has requirements for intelligent and autonomous systems in future missions that can benefit from advances in situation management. We describe requirements for the integrated vehicle health management program where our biology-inspired architecture provides a layered approach and decisions can be made at the proper level to improve safety, reduce costs, and improve efficiency in making diagnostic and prognostic assessments of the structural integrity, aerodynamic characteristics, and operation of aircraft

Situation management is a rapidly evolving science where managed sources are processed as realtime streams of events and fused in a way that maximizes comprehension, thus enabling better decisions for action. Sensor networks provide a new technology that promises ubiquitous input and action throughout an environment, which can substantially improve information available to the process. Here we describe a program of NASA that requires improvements in sensor networks and situation management. We present an approach for massively deployed sensor networks that does not rely on centralized control but is founded in lessons learned from the way biological ecosystems are organized. In this approach, fully distributed data aggregation and integration can be performed in a scalable fashion where individual motes operate based on local information, making local decisions that achieve globally-meaningful effects. This exemplifies the robust, fault-tolerant infrastructure required for successful situation management systems

Research in wireless sensor network technology has exploded in the last decade. Promises of complex and ubiquitous control of the physical environment by these networks open avenues for new kinds of science and business. Due to the small size and low cost of sensor devices, visionaries promise systems enabled by deployment of massive numbers of sensors working in concert. Although the reduction in size has been phenomenal it results in severe limitations on the computing, communicating, and power capabilities of these devices. Under these constraints, research efforts have concentrated on developing techniques for performing relatively simple tasks with minimal energy expense assuming some form of centralized control. Unfortunately, centralized control does not scale to massive size networks and execution of simple tasks in sparsely populated networks will not lead to the sophisticated applications predicted. These must be enabled by new techniques dependent on local and autonomous cooperation between sensors to effect global functions. As a step in that direction, in this work we detail a technique whereby a large population of sensors can attain a global goal using only local information and by making only local decisions without any form of centralized control.

With the advent of Micro-Electrical-Mechanical Systems (MEMSs) technology, complex and ubiquitous control of the physical environment by machines will facilitate the diversity of mechanization and automation, long promised by visionaries. Wireless sensor networks have appeared as the first generation of this revolutionary technology. The small size and low cost of sensor devices will enable deployment of massive numbers, but initially place severe limitations on the computing, communicating, and power capabilities of these devices. With these constraints, research efforts have concentrated on developing techniques for executing simple tasks with minimal energy expense. But, as MEMS evolve, computing and communicating capabilities are expected to improve at an accelerating rate and new techniques for supplying energy will significantly reduce the low power constraint. Increased capabilities will be possible, and it is predicted that societies of machines will evolve to be autonomous, cooperative, fault-tolerant, selfregulating, and self-healing. Improvements in biomimetic software and evolvable hardware will lead to self-sustaining communities of machines with emergent behavior that autonomously operate and adapt to changes in the environment. The main goal of this chapter is to investigate biomimetic models in relation to their potential application to the evolution of these systems, thus providing a framework that guidesCHAPMAN: "C4754_C033" — 2005/8/6 — 14:13 — page 602 — #2the evolution from the current primitive organizations of sensor nodes to pervasive societies of intelligent electromechanical systems.

Promises of ubiquitous control of the physical environment by massively-deployed wireless sensor networks open avenues for new applications that will redefine the way we live and work. Due to small size and low cost of sensor devices, visionaries promise systems enabled by deployment of massive numbers of sensors ubiquitous throughout our environment working in concert. Recent research has concentrated on developing techniques for performing relatively simple tasks with minimal energy expense, assuming some form of centralized control. Unfortunately, centralized control is not conducive to parallel activities and does not scale to massive size networks. Execution of simple tasks in sparse networks will not lead to the sophisticated applications predicted. We propose a new way of looking at massively-deployed sensor networks, motivated by lessons learned from the way biological ecosystems are organized. We demonstrate that in such a model, fully distributed data aggregation can be performed in a scalable fashion in massively deployed sensor networks, where motes operate on local information, making local decisions that are aggregated across the network to achieve globally-meaningful effects. We show that such architectures may be used to facilitate communication and synchronization in a fault-tolerant manner, while balancing workload and required energy expenditure throughout the network.

Research in wireless sensor network technology has accelerated rapidly in the last decade. Promises of ubiquitous control of the physical environment by these networks open avenues for new applications that will redefine the way we live and work. Due to the small size and low cost of sensor devices, visionaries promise systems enabled by deployment of massive numbers of sensors working in concert. Recent sensor network research has concentrated on developing techniques for performing relatively simple tasks with minimal energy expense, assuming some form of centralized control Unfortunately, centralized control does not scale to massive size networks and execution of simple tasks in sparse networks will not lead to the sophisticated applications predicted. The contribution of this work is to continue our work in looking at massively-deployed sensor networks, motivated by lessons learned from the way biological ecosystems are organized. We demonstrate that in such a model, fully distributed synchronization can be performed in a scalable fashion in massively deployed sensor networks, where individual motes operate based on local information, making local decisions that are aggregated across the network to achieve a globally-meaningful effect.

Promises of ubiquitous control of the physical environment by large-scale wireless sensor networks open avenues for new applications that are expected to redefine the way we live and work. Most of recent research has concentrated on developing techniques for performing relatively simple tasks in small-scale sensor networks assuming some form of centralized control. The main contribution of this work is to propose a new way of looking at large-scale sensor networks, motivated by lessons learned from the way biological ecosystems are organized. Indeed, we believe that techniques used in small-scale sensor networks are not likely to scale to large networks; that such large-scale networks must be viewed as an ecosystem in which the sensors/effectors are organisms whose autonomous actions, based on local information, combine in a communal way to produce global results. As an example of a useful function, we demonstrate that fully distributed consensus can be attained in a scalable fashion in massively deployed sensor networks where individual motes operate based on local information, making local decisions that are aggregated across the network to achieve globally-meaningful effects.

Promises of ubiquitous control of the physical environment by massively-deployed wireless sensor networks open avenues for new applications in support of situation management. Recent sensor network research has concentrated on developing techniques for performing relatively simple tasks with minimal energy expense, assuming some form of centralized control. Unfortunately, centralized control is not conducive to situation management as it allows single points of failure and does not scale to massive size networks. We propose a new way of looking at massively-deployed sensor networks, motivated by lessons learned from the way biological ecosystems are organized. We demonstrate that in our model fully distributed data aggregation and integration can be performed in a scalable fashion where individual motes operate based on local information, making local decisions that are aggregated across the network to achieve globally-meaningful effects. This exemplifies the robust, fault-tolerant infrastructure required for successful situation management systems

Promises of ubiquitous control of the physical environment by sensor networks open avenues that will redefine the way we live and work. Due to the small size and low cost of sensors, visionaries promise smart systems enabled by deployment of huge numbers of sensors working in concert. At the moment, sensor network research is concentrating on developing techniques for performing simple tasks with minimal energy expense, assuming some form of centralized control. Centralized control does not scale to large networks and simple tasks in small-scale networks will not lead to the sophisticated applications predicted. Recently, the authors have proposed a new way of looking at sensor networks, motivated by lessons learned from the way biological ecosystems are organized. Here we demonstrate that in such a model, fully distributed data aggregation can be performed efficiently, without synchronization, in a scalable fashion, where individual motes operate autonomously based on local information, cooperating with neighbors to make local decisions that are aggregated across the network achieving globally-meaningful effects.

The natural world may be the inspiration we need for solving our computer problems. While it is certainly true that "the map is not the territory," most visitors to a foreign country do prefer to take with them at least a guidebook to help locate themselves as they begin their explorations. That is the intent of this article. Although there will not be enough time to visit all the major tourist sites, with a little effort and using the information in the article as signposts, the intrepid explorer can easily find numerous other, interesting paths to explore.

The defining characteristic of a multicellular organism is unity of purpose. In biology, the purpose is survival of the organism. The purpose of our multi-agent system is to provide a persistent computing environment in harsh conditions where repairs are difficult, or impossible. The multi-agent organism is a single entity built from logically dependent cells, where each cell is a discrete, independent hardware-processing unit. Similar to biology, each cell contains a full description of the system encoded as genes in a software genome. Cells choose which gene to express depending on internal state, the genome, and the state of neighboring cells. Gene expression involves executing a program fragment, which, when combined with all other genes in the genome, defines the full system. The multi-agent architecture provides a computing environment that adapts to unexpected changes in the hardware by reconfiguring to the new hardware without losing functionality, although performance may be affected.