This Small Business Innovation Research (SBIR) Phase I project will develop a new capability for robust system identification that will enable more effective neurocontrol. This research will specifically address control of complex nonlinear dynamic systems. It will develop figures-of-merit for system identification, useful to translate quality-evaluation metrics into terms that can be readily used for computation of a strategic utility function in an adaptive-critic neurocontrol architecture. It will explore biologically-based approaches to system identification that will permit modeling of more complex systems, including systems which evolve or change over time. It will also enhance cabilities to build system identification with complex temporal associations, allowing more complex processes to be controlled. One or more of these innovations will be hosted in an adaptive critic architecture. Feasibility demonstrations will select and use an appropriate function for optimization and use of context-dependent processes in producing a system model. The improved system identification methods developed in this project will, when inserted into a robust neurocontrol system such as an adaptive critic, permit adaptive control of a wide range of systems with applications to flight control, control of electric cars, control of biomedical devices, robotic motion control, and many others. This capability will be applied to a currently developing commercial flight control technology in hypersonic waverider as well as potentially to the next-generation `clean car.`
