Circumvention of Pedestrians While Walking in Virtual and Physical Environments

Abstract

Virtual environments (VEs) are increasingly used in the context of scientific inquiries and rehabilitation for tasks that are otherwise difficult to control or perform safely in physical environments (PEs), such as avoiding other pedestrians during locomotion. The usefulness of VEs, however, remains constrained by the extent to which they can elicit natural responses. The objectives of the study were to examine circumvention strategies in response to pedestrians approaching from different directions in the VE versus PE and to determine the effects of repeated practice on the circumvention strategies. Twelve participants were assessed over five blocks of eight trials that consisted of walking toward a target while circumventing pedestrians approaching from different directions (0°, ± 30° right or left or none) in the VE and the PE. Similar onset distances of circumvention strategy and preferred side of circumvention were observed between the two environments. Participants, however, maintained enlarged minimum distances from the interferer (13%) and walked slower (11.5%) in the VE. Repeated practice resulted in walking speed increments of 7.4% over the entire session that were similar in the VE versus PE. While the changes observed in VE may reflect the use of more cautious circumvention strategies, the similarities in strategies between the two environments and the advantages of VEs (e.g., controlled exposure, reproduction of ecologically valid conditions, and safety) suggest that virtual reality is a valuable tool to study visually guided locomotor tasks, such as pedestrian circumvention, and shows great potential for assessment and intervention in physical rehabilitation.

Studying Gaze Behaviour during Collision Avoidance with a Virtual Walker: Influence of the Virtual Reality Setup

“…We found a difference in the average walking speed that was lower in the HMD-W condition: participants wore a HMD but had to move among physical obstacles, which may have induced a safer locomotion speed. This result is also consistent with previous works [1,6].…”

“…i.e., mpd variations and not absolute values, the absence of significant difference between the studied conditions suggests that motion adjustments are performed similarly in both real and virtual environments. This corroborates results from previous studies [6,14,31]. Furthermore, considering the VR conditions only, we did not find any significant difference between mpd values along the whole collision avoidance task.…”

“…The increasing interest in VR tools to perform human experiments stresses the need for validating them, i.e., to compare human behaviour in VR and reality. Several works already performed this comparison for locomotion studies, either considering goal directed [1,10] or collision avoidance tasks [1,6,14,31]. It was shown that virtual and real trajectories have common properties, even if some quantitative differences remain such as slower walking speeds or increased clearance distances.…”

Toward Virtual Reality-based Evaluation of Robot Navigation among People

“…Moreover, the HMD we used has a limited field of view which impacts the participants immersion. Concerning the pilot experiment, the number of participants is low (5). However, it is enough to make a proof of concept and to identify ways of improvement.…”

“…Then, comparing same tasks performed in VR and in real conditions is required to estimate those biases and avoid misinterpretations of experimental results. Such comparisons have been done for many different studies: locomotion toward a goal [1,8], collision avoidance [1,5,16,30], personal space [19,28], etc. Also, HRI experiments in VR in [23] explore the bias when the human only is immersed in VR.…”