Time at the curb, not the rate of head scanning, distinguishes street crossing in virtual reality with severe-to-profound visual impairment

Purpose: We tested whether people with severe-to-profound visual impairment compensate at street crossings with more head scanning and/or more time at the curb. Methods: Ten normally sighted controls, tested with normal vision and with Bangerter filters simulating severe-to-profound visual impairment (approximately 20/2000), and 17 adults with severe-to-profound visual impairment crossed a virtual street in a head-mounted display. Four complexity levels ranged from an empty street to yielding vehicles with pedestrian avatars; only the two most complex levels contained moving vehicles. Cumulative absolute head yaw and latency to step-off were analyzed with mixed-effects models, and latencies were fitted with an ex-Gaussian distribution. Results: Compared with normal vision, latency to step-off was 1.57 times longer with simulated and 1.82 times longer with clinical severe-to-profound visual impairment (Bonferroni p < 0.001 and p = 0.008, respectively). In severe-to-profound visual impairment, the additional latency was attributable to the slow tail of the latency distribution (exponential component τ longer by 3.40 s without and 6.20 s with moving vehicles; Bonferroni p = 0.037 and 0.022, respectively); peak position did not differ significantly. Before step-off, cumulative absolute head yaw increased with complexity, F(3, 115.8) = 37.16, p < 0.001, and was 1.88 times higher with simulated impairment than with normal vision (Bonferroni p = 0.024); it did not differ significantly between clinical impairment and normal vision, either before step-off or during walking (both Bonferroni p ≥ 0.479). During walking, its increase with complexity depended on viewing condition, F(6, 109.3) = 5.23, p < 0.001. Conclusions: Severe-to-profound visual impairment was associated with longer and more variable time at the curb, without a significant increase in cumulative head scanning, although the increase in head scanning with complexity during walking was attenuated. Time at the curb is relevant to orientation and mobility practice and accessible crossing design.