Automated manual training using adaptive neuro-controllers

This paper presents an application of artificial neural networks in automated manual training. The idea of an automated manual trainer is to train a human to operate a complex system automatically with limited human interaction. An automated helicopter training system was used to demonstrate the feasibility of this study. Neuro-controllers in conjunction with critic models are used to implement the automated helicopter trainer to perform hover maneuvers. The simulation results indicate that the automated helicopter trainer is capable of performing the hover maneuvers. To verify the concept of the automated trainer, a simple one axis manual control cart-pole system was used. Two groups of four students each were used in the experiment with one being the experimental group which was trained with the automated trainer, and the other being the control group which was trained with no automated trainer. The experimental results show that the automated manual trainer did not “mess up” the training process, but no further conclusions can be drawn since the experimental group is too small to generalize the results. Further study in this area is recommended