Evolvable Systems (ES)

Project Manager: Jason Lohn, NASA Ames Research Center

The goal of the ES project is to dramatically increase mission survivability and science return through the development and application of biologically inspired, adaptive, autonomous systems. To reach this goal, ES will research and develop evolutionary and adaptive algorithms in three main focus areas:

the reconfiguration and re-use of available resources for new or existing functions
the optimization of existing systems and the design of new systems using techniques such as evolutionary algorithms
the adaptation of systems to new environments through sensing, learning and automated discovery

ES's principal customers are the Space Science Enterprise (Code S), the Biological & Physical Research Enterprise (Code U), and the Earth Science Enterprise (Code Y).

ES includes three research elements:

Dynamic Evolution for Fault Tolerance

This element is investigating methods for making electronic systems fault tolerant. The ability of these systems to adapt and recover from failure during NASA missions is integral to achieving the agency's top priority of mission safety. One category of failures concerns single event upsets and latchups to onboard electronics. For example, in high-radiation environments, the cumulative effects of charged particles can damage spacecraft electronics, even with extensive shielding. The underlying premise to this research is that the healthy resources that remain after a failure can be commandeered to provide lost functionality.

Evolutionary Algorithms for Scheduling

This element is investigating the use of evolutionary algorithms for NASA scheduling problems. Examples include satellite, spacecraft, and telescope scheduling; logistics support; shuttle operations; and robotic autonomy. These problems can be approached by a variety of automated techniques, including those from operations research and artificial intelligence. Given a new NASA scheduling application, this element's research will provide a way to determine which search techniques are likely to perform well, and how much time or effort the solutions will require.

Evolvable Hardware for Sensors

This element is investigating new capabilities in space electronics, especially space sensing systems. These include the ability of these systems to intelligently reconfigure themselves to adapt to new and/or changing mission requirements and environments. The element's objectives are to demonstrate the automated synthesis of signal processing functions used in space applications, and to demonstrate increased survivability in the presence of faults and extreme environments. Current research focuses on in-situ evolution of custom-made, evolution-oriented field-programmable transistor arrays.

ES Project Milestones

Demonstration of fault recovery using direct, on-the-chip hardware evolution. (Project milestone 8.3)
Demonstration of ability to recover functionality from injected programmable logic fault. (Project milestone 8.3.1)
Demonstration of a design and a comparative analysis of evolutionary algorithms in Earth-observing satellite scheduling problems. (Project milestone 8.3.2)
Development and demonstration of improved electronic sensor systems, which can perform without faults above 250 degrees. (Project milestone 8.3.3)
Design of novel devices, architectures and algorithms for evolutionary and adaptive behavior in nanoelectronics. (Project milestone 8.13)
Demonstration of evolutionary and adaptive algorithms tailored for searching the space of nanoelectronic architectures. (Project milestone 8.13.2)

Top