Modeling Human Multimodal Perception and Control Using Genetic Maximum Likelihood Estimation

Abstract

This paper presents a new method for estimating the parameters of multi-channel pilot models that is based on maximum likelihood estimation. To cope with the inherent nonlinearity of this optimization problem, the gradient-based Gauss-Newton algorithm commonly used to optimize the likelihood function in terms of output error is complemented with a genetic algorithm. This significantly increases the probability of finding the global optimum of the optimization problem. The genetic maximum likelihood method is successfully applied to data from a recent human-inthe-loop experiment. Accurate estimates of the pilot model parameters and the remnant characteristics were obtained. Multiple simulations with increasing levels of pilot remnant were performed, using the set of parameters found from the experimental data, to investigate how the accuracy of the parameter estimate is affected by increasing remnant. It is shown that only for very high levels of pilot remnant the bias in the parameter estimates

Effects of Peripheral Visual Cues in Simulator-Based Training of Multimodal Control Skills

“…In order to identify and model the multi-channel human operator response, characterized by H pe and H p φ , the disturbance signal f d and the target signal f d were independent sum-of-sines signals. [17][18][19] Given the quasilinear human operator model used, the control input had contributions from the error response, u e , the roll response, u φ , and a remnant n accounting for nonlinear behavior and measurement noise.…”

“…(3) to (5) contained seven free parameters (K e , T lead , τ e , K φ , τ φ , ω nm , and ζ nm ), which were estimated from the collected experimental data (the time-domain signals e, φ, and u) using the time-domain parameter estimation technique of Ref. 19. For the training phase data of Group NV only H pe was fitted, as no out-of-the-window visual or motion cues were available.…”

“…Values for the VAF in tracking experiments are usually around 80% and 90% when human operator data is averaged between repeated measurements. 19 In a training experiment, the evolution in the model throughout the runs is crucial to evaluate the acquisition and development of control skills and averaging results would mask that effect. Therefore, a model was fit to each individual run without any averaging, lowering its VAF due to higher remnant noise.…”

Estimation of Time-Varying Pilot Model Parameters

“…[1][2][3][4] In order for these techniques to provide accurate identification results, specific requirements are posed on the design of experiments in which human control behavior is estimated. These requirements ensure that the human controller behaves like a stationary control element in the control loop.…”

“…3 A technique exists to estimate the model parameters from time-domain data in a single step using maximum likelihood estimation (MLE), 4 significantly reducing bias and variance compared to the two-step methods. However, this technique has not been used for the estimation of time-varying parameters in the field of manual control.…”