KSEF RDE: Assistive Learning for Human-Machine-Interaction Systems

Humans learn to interact with a variety of complex systems. However, learning to interact with complex systems can be challenging, and there are many cases where enhancing the learning process would provide significant benefits. The objective of this research effort is to develop techniques of automatic assistive learning (AAL) to enhance human motor learning. Our research approach utilizes the findings from our previous human-learning experiments, and is based on two hypotheses: (H1) Humans attempt to learn the dynamics of the systems with which they interact and use the inverse dynamics in feedforward. (H2) A human's learning rate can be improved by allowing them to interact initially with a simple dynamic system and by making the system incrementally more complex until the human is interacting with the final desired system. Hypothesis H1 is a specialization of the internal model hypothesis, which is the predominant neuroscience theory of human motor learning and has been tested extensively in the literature. We propose two phases of research. Phase 1 uses experiments with human subjects to test H2. Phase 2 seeks to develop AAL techniques based on H1 and H2 and to test those techniques on experiments with human subjects. This project seeks to establish a new direction in assistive learning research, which has been confined primarily to the medical fields. This project will identify beneficial human learning mechanisms and produce techniques that enhance a human's ability to learn to interact with complex dynamic systems. We believe that this project will be the first effort in a new area of dynamic systems research. The assistive learning techniques that result from this project will promote innovation in human interface technologies. Specific applications of this research include rehabilitation technologies, orthopedic devices, and various human training programs.