HUMAN VS FILTER AS DATA EXTRAPOLATOR IN A TWO-COORDINATE, SAMPLED-DATA TRACKING SYSTEM.
Abstract
The performance of human operators was compared with that of a singly augmented filter in the continuous determination of the present position of a constant rate target moving in two coordinates. Target position was indicated intermittently in low, medium, or high noise levels and at low, medium, or high data rates. In addition, the target was subjected to a 10, 20, or 60 degrees course change in each trial. The filter evidenced less average tracking error in 23 of the 27 combinations of conditions of data rate, noise level, and course change. In twelve of these instances the filter was significantly superior at p = .02 level. Also, the results indicated increased error in human and filter performance as a function of increasing noise levels and decreasing data rates. Relative to the further enhancement of data extrapolation, several avenues of investigation recommend themselves. An immediate possibility is the employment of filter networks as an aid to the human operator. A second avenue of investigation is the study of more sophisticated filter designs. This experiment employed a filter of fixed time constant and fixed augmentation. An optimum filter would be of an 'adaptive' type, automatically adjusting its time constant and augmentation as a function of noise and data rate. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 14, 1965
- Accession Number
- AD0624494
Entities
People
- C. L. Tipton
Organizations
- United States Naval Research Laboratory