Abstract
Previous studies have clearly demonstrated the benefit of providing the human operator with input preview information during manual control tasks. This study was designed to analyse if and to what extent such preview benefit depends on whether the preview extends immediately ahead of the ‘ present’ position or is postponed (lagged) by a certain amount. The experimental results from eight subjects performing a purely digital pursuit control task with a first-order controlled system showed that the performance deteriorated (RMS error increased) nearly linearly with increasing preview lag for a band-limited white Gaussian noise input. However for a first-order autoregressivc process (with parameter α =0·9) as reference input, the control performance was clearly less affected by the preview lag. Irrespective of the extent of the preview lag, performance was clearly superior for α equals; 0·9 as compared to α =o,