Effects of Preview on Human Control Behavior in Tracking Tasks With Various Controlled Elements

Abstract

This paper investigates how humans use a previewed target trajectory for control in tracking tasks with various controlled element dynamics. The human's hypothesized "near" and "far" control mechanisms are first analyzed offline in simulations with a quasi-linear model. Second, human control behavior is quantified by fitting the same model to measurements from a human-in-the-loop experiment, where subjects tracked identical target trajectories with a pursuit and a preview display, each with gain, single-, and double-integrator controlled element dynamics. Results show that target-tracking performance improves with preview, primarily due to the far-viewpoint response, which allows humans to cancel their own and the controlled element's lags, without additional control activity. The near-viewpoint response yields better target tracking at higher frequencies, but requires substantially more control activity. The control-theoretic approach adopted in this paper provides unique quantitative insights into human use of preview, which can help to explain human behavior observed in other preview control tasks, like driving.

Effects of Preview Time in Manual Tracking Tasks

“…HC use of the full previewed target is modeled with two responses. The main, low-frequency, "far-viewpoint" response allows for most performance improvement relative to zero-preview, pursuit tasks, while the auxiliary, highfrequency, "near-viewpoint" response improves performance slightly more at a cost of a substantial increase in control effort [14]. The far-viewpoint response involves an identical feedback control-strategy as in compensatory tracking tasks [12]; however, the error e ⋆ (t) that is minimized is not the true error e(t), but the difference between the filtered target at the far viewpoint τ f s ahead and the CE output:…”

“…HCs completely ignore the target signal variations when K f =0, while K f =1 indicates a response to the true advanced error, at least at those frequencies sufficiently below the smoothing filter cutoff 1/T l, f [14]. In the time domain, the low-pass filter in (1) can be interpreted as the weighted average of the previewed target up to the far viewpoint, such that τ f is the most distance point on the previewed trajectory that is used by the HC, while T l, f quantifies the portion of the visible preview that is used for smoothin the trajectory.…”

“…Task characteristics (or variables) such as the Controlled Element (CE) dynamics and target trajectory bandwidth also evoke HC behavior adaptations [12], [13], and influence the effects of preview, including the critical time [5], [14]. Attempts to quantify the effects of preview time through control-theoretic modeling of driver steering [1], [15]- [18] and manual tracking [2], [3], [5], [19] account for HCs' use of preview in fundamentally different ways.…”

“…Preliminary experimental results suggest that this model can also capture HC adaptation to preview time [7], but this was only shown for rate control tasks with preview times between 0 and 1 s. The model's physically interpretable parameters can be directly estimated from experimental data using system identification, yielding novel quantitative insights into how HCs use preview. This already led to an initial set of "verbal adjustment rules" that cover HC adaptation to the CE dynamics in preview tracking tasks [14]. Unfortunately, we currently lack the understanding and experimental data that are required to formulate similar rules for the effects of other task variables, such as available preview time [13].…”

“…As the CE dynamics are known to strongly affect HCs' use of preview [14], we investigate the effects of preview time with both Single Integrator (SI, rate control) and Double Integrator (DI, acceleration control) CE dynamics.…”