Walter F. Truszkowski

Neural Basis Function Control of Super Micro Autonomous Reconfigurable Technology (SMART) Nano-Systems

** Nanotechnology, taken to its full three-dimensional potential, will place within the volume of a cube of sugar systems of vast complexity that far exceed the quantitative and qualitative capabilities of today’s largest supercomputers. Currently, the uncertainty and imprecision of the real world is tamed, rigidly fixed, by addressable, digital techniques and the careful orchestration of digital patterns within our machines. How to handle the interaction between our digitally implemented systems and continuous, disorganized nature is a key question. NASA is currently researching ways to move beyond autonomy implemented as bruteforce control over every degree of freedom we can discover in our systems. Our systems operate in natural environments: inhumanly harsh, unfamiliar, unknown, and uncontrolled environments. Nature often surprises us, and so we turn to natural systems for clues about how to make massively complex systems more robust, reliable, and truly autonomous. Turning to Computer Science we draw on what we’ve learned about multi-agent systems running continuously and autonomously to understand information flow at the highest semantic levels. From physics we recall that the behaviors of systems may often be enumerated in a basis of fundamental behaviors. Non-linear physics contains clues about how to connect the physical world with the patterns of electric signals that make up the soft, information component of the systems. Genetics and control theory instruct how to handle long and short-term feedbacks throughout the system. Chemistry and biology provide important guiding principles governing system functions.

Intelligent Systems in the Evolvable ANTS Architecture

The Autonomous Nano Technology Swarm (ANTS) is a breakthrough in mission architecture that enables distributed multi-platform missions to explore autonomously multiple-faceted targets. ANTS is a biologically inspired system architecture that is adaptable, reconfigurable and evolvable in all its system hierarchies, from the swarm level down to the subsystem and component level. Both the hardware and software that constitute each layer in the ANTS architecture and their interfaces are evolvable and adaptable. Advanced autonomy is crucial in the operations of ANTS systems. We recognize the sheer computational power required by the Synthetic Neural System that is central to ANTS and have begun efforts to attain space-qualified high-performance computing capabilities. I. Introduction he Autonomous Nano Technology Swarm (ANTS) is a breakthrough in mission architecture that enables distributed multi-platform missions to explore autonomously multiple-faceted targets. These exploration missions operate in a discovery mode, so a rigid architecture with a pre-determined mission plan would not be able to respond well to unexpected findings and to exploit targets of opportunity. ANTS is a biologically inspired system architecture that is adaptable, reconfigurable and evolvable in all its system hierarchies, from the swarm level down to the subsystem and component level. Both the hardware and software that constitute each layer in the ANTS architecture and their interfaces are evolvable and adaptable. The large reconfigurable gossamer space frames are built on the same self-similar architecture as the MEMS-based reconfigurable flight electronics. A self-similar Synthetic Neural System controls each system node and the interfaces, integrating their heuristic functions and autonomic tasks. Many technology developments are required for the implementation of the ANTS architecture, including nanotechnology, advanced materials, miniaturized system components, and intelligent systems. Advanced autonomy is crucial in the operations of ANTS systems. We recognize the sheer computational power required by the Synthetic Neural System that is central to ANTS and have begun efforts to attain space-qualified high-performance computing capabilities. We are developing a validation experiment using multiple COTS Von Neumann processors for space-based Beowulf cluster in an effort funded by the NASA Space Technology 8 program. In addition, we investigate the use of Reconfigurable Data Path Processors in a non-Von Neumann architecture to increase the onboard data analysis capability and to enable in-situ knowledge discovery.