This work describes a thermodynamically motivated neural network model that self-organizes to transport charge associated with internal and external potentials while in contact with a thermal bath. Isolated networks show multiscale dynamics and evidence of phase transitions, and externally driven networks evolve to efficiently connect external positive and negative potentials. The model implements techniques for rapid, global, reversible, conservative equilibration of node states followed by slow, local, irreversible, dissipative adaptation of the edge states as a means to create multiscale order. All interactions in the network are local and the network structures can be generic and recurrent. The model integrates concepts of fluctuation, dissipation, adaptation and equilibration and offers an illustration of the thermodynamic evolution of organization in open systems. The key conclusion of the work is that the dissipation of conserved physical quantities can drive the self-organization of open thermodynamic systems.

In October 2005 the Defense Advanced Research Projects Agency (DARPA) launched the Nano Air Vehicle (NAV) program with the objective of developing and demonstrating small (<10 cm wingspan), lightweight (<10 grams) air vehicle systems with the potential to perform challenging indoor and outdoor military missions. 1 The program was executed in two phases. The initial 18-month phase included four contractors-AeroVironment Inc., Charles Stark Draper Laboratory, Lockheed Martin Advanced Technology Laboratories, and MicroPropulsion Corp.-each creating a preliminary design of its NAV system through a combination of analysis, component testing, trade studies, and concept development.Technologies developed in Phase 1 included rotating and flapping wings for lift generation and vehicle control, algorithms for stability and control using video images, and analytical tools for low Reynolds number aircraft. In Phase 2 AeroVironment was selected to continue the development begun in Phase 1. Phase 2 concluded with a hummingbird-inspired prototype aircraft with two flapping wings that for the first time demonstrated precision hovering and fast forward flight using only the flapping wings for propulsion and control. This paper will describe the initial program goals, outline the technological challenges addressed, briefly describe the individual NAV projects and related technology programs, and provide insight into technical areas still in need of further development. I. NAV Program Motivation and GoalsThe military forces of the United States and its allies have an ever-present need for improved intelligence, surveillance, and reconnaissance (ISR). Detailed ISR on the ground and in urban environments is a particular highvalue challenge that can be viewed as a gap in current ISR capabilities. Warfighters currently lack a technology that can be readily and flexibly deployed as a mobile sensing system within buildings, around corners, over walls, and in other denied areas prior to entry. Appropriately instrumented nano air vehicles (NAVs), unmanned aerial vehicles (UAVs) approximately the size of large flying insects or small birds, could provide such mobile sensing capability, as well as persistent sensing through the delivery of small sensor packages. NAVs would expand and complement the current capabilities provided by ground (wheeled and tracked) robots and thrown sensors, because these devices are generally limited in mobility and perspective inside buildings and can be considerably larger and heavier than a NAV-sized aircraft.

No abstract

Concepts from thermodynamics are ubiquitous in computing systems today—e.g., in power supplies and cooling systems, in signal transport losses, in device fabrication, in state changes, and in the methods of machine learning. Here we propose that thermodynamics should be the central, unifying concept in future computing systems. In particular, we suppose that future computing technologies will thermodynamically evolve in response to electrical and information potential in their environment and, therefore, address the central challenges of energy efficiency and self-organization in technological systems. In this article, we summarize the motivation for a new computing paradigm grounded in thermodynamics and articulate a vision for such future systems.

Mental imagery and planning are important aspects of cognitive behaviour. Being able to predict outcomes through mental simulation can increase environmental fitness and reduce uncertainty. Such predictions reduce surprise and fit with thermodynamically driven theories of brain function by attempting to reduce entropy. In the present work, the authors tested these ideas in a predator-prey scenario where agents with a limited energy budget had to maximise food intake, while avoiding a predator. Forward planning agents, with the ability to mentalise, to Actor Critic agents that do not plan beyond the current state were also compared. The authors show that the ability to mentalise has distinct advantages when in noisy, uncertain stimuli. These advantages are even more prevalent when tested in the real world on physical robots. The authors' results highlight the importance of taking into consideration mental imagery and embodiment when constructing artificial cognitive systems.