Purchase this article with an account.
Li Li, Leland Stone, Erich Chan; Visual control of steering toward a goal uses heading but not path information. Journal of Vision 2008;8(6):1162. doi: https://doi.org/10.1167/8.6.1162.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Instantaneous direction of self-translation (heading) and trajectory (path) are two central features for the control of locomotion. We have shown that humans can perceive heading in the absence of path information (Li, Sweet, & Stone, JOV 2006). Here we investigate whether humans use a heading- or path-based control strategy when steering a vehicle toward a goal. We simulated locomotion over a ground plane at 2 m/s with observers' line of sight fixed with respect to the vehicle and randomly offset by ±8° to render screen-centering strategies useless. Observers used a joystick to change the curvature of the vehicle's traveling path to steer toward an environmental target under two display conditions (sparse flow: the ground was composed of random dots; dense-flow: the ground was textured). In the dense-flow condition, we examined both open- and closed-loop performance (open-loop: the target disappeared as observers initiated steering; closed-loop: the target was visible throughout the trial) while in the sparse condition, we only examined the former. For 7 observers (5 naïve), in all conditions, the maximum path curvature was several times larger than that expected if observers used a path-based strategy to point their expected path at the target and then hold curvature constant (mean curvature ratio±SE: 15.1±4.7, 9.1±1.7, & 17.3±8.2, for sparse open-loop, dense open-loop and closed-loop, respectively). Instead, observers over-steered initially and then let path curvature decrease toward zero over time with the final heading error larger for sparse-flow than for either open- or closed-loop dense-flow (mean±SE: 5.0±1.1°, 1.2±0.3°, & 1.7±0.4°, respectively). This behavior is consistent with observers steering their heading (estimated from optic flow) towards the goal. The fact that final heading error is indistinguishable in the open- and closed-loop conditions argues against a Tau-equalization strategy. Humans rely primarily on heading when steering toward a goal and not on estimated future path or time-to-contact information.
This PDF is available to Subscribers Only