The objective of this project is to define a science which allows for the evolution of complex software systems that can fully utilize massively parallel computers of ever-greater capability (initially dual quad cores running fewer higher-level sensors for tractability) and employ this design in human-robot interaction with a final goal of full autonomy of the robot (in this case, an unmanned helicopter). In order to test a new learning approach that can bestow intelligence on a system, based upon human (tutor) – robot (student) interaction, we are working jointly to design an intelligent system architecture, based on the KASER (Knowledge Amplification by Structural Expert Randomization), and plan to implement the methodology on an aerial vehicle (a helicopter), which will execute several maneuvers such as basic hovering, steep approach, confined area approach, and basic altitude flying. The software to perform these maneuvers will be developed using two methods (teach mode with a human-in-the-loop) and classical control. Then, the knowledge amplification system will autonomously generate another set of software, which will then be tested and compared to the execution of the baseline two software codes using the same helicopter flight scenarios. We anticipate that such complex functional real-time software can be most cost-effectively written through the use of a software-writing system embodying knowledge amplification.