Virtual Navigation for Blind People: Building Sequential Representations of the Real-World
João Guerreiro, Dragan Ahmetovic, Kris M. Kitani, Chieko Asakawa · 2017 · Proceedings of the 19th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '17) · doi:10.1145/3132525.3132545
Summary
This paper investigates whether blind people can build accurate mental representations of real-world routes by virtually navigating them on a smartphone before physically visiting the location. Sighted people routinely study maps before visiting new places, building a sequential mental model of turns, distances, and landmarks. Blind people lack an equivalent pre-navigation tool. The researchers built on the NavCog indoor navigation system to create two smartphone-based virtual navigation interfaces. VirtualWalk simulates walking by detecting phone tilt via gyroscope — tilting the phone forward triggers auditory footstep sounds at variable speeds (0.3-0.7 seconds per 0.7-meter step), while rotating the phone performs turns. VirtualLeap uses swipe gestures to jump between intersections and points of interest (POIs) along the route, announcing distances, turn directions, and nearby POIs at each point. Both interfaces announce POI names and their relative position (e.g., "Starbucks is on your right") as the user passes them. Route data was automatically generated from OpenStreetMap with POIs sourced from Yelp and FourSquare. The system places POIs in their functional position within the route sequence rather than using Euclidean distance, allowing them to serve as memorable sequential landmarks.
Key findings
In a study with 14 visually impaired participants (ages 41-75, mean 59.15; 13 legally blind, 1 low vision), participants used both methods on 130-meter routes with 4 intersections and 5 POIs. Route reconstruction using LEGO blocks showed most participants accurately learned the sequential structure: 10 of 13 reconstructed the route form correctly with VirtualWalk and 8 with VirtualLeap. Participants placed an average of 4.54 (VirtualLeap) and 4.69 (VirtualWalk) of 5 POIs, with most placed in the correct street block. However, estimating relative block lengths was difficult for both methods. In a real-world exposure task at Carnegie Mellon University, all participants reached the destination after virtually navigating the route, with 7 of 13 making zero navigation errors. Mean route accuracy was 0.82. POIs served as essential landmarks — participants used them to provide accurate turn instructions and to recover from errors by returning to a known POI. Both methods produced similar spatial understanding with no significant differences, but participants preferred VirtualWalk for building detailed cognitive maps and VirtualLeap for quick route overviews and POI exploration. VirtualLeap took significantly longer (mean 540s vs 413s) as users explored routes more times (9.15 vs 5.46 iterations).
Relevance
This research addresses a fundamental equity gap in navigation: sighted people take for granted the ability to preview a route on a map before visiting, while blind people typically arrive at new locations with no spatial understanding. The ability to build a mental representation beforehand increases confidence, independence, and safety. The finding that POIs serve as sequential landmarks — enabling both accurate navigation instructions and error recovery — has practical implications for how navigation apps should present route information to blind users. Rather than treating POIs as supplementary exploration data, they should be integrated as core structural elements of route guidance. For accessibility practitioners, the complementary strengths of the two modalities suggest that navigation tools should support multiple exploration strategies: quick overview scanning (VirtualLeap) and detailed spatial immersion (VirtualWalk). Participants' requests for transit information, intersection details, crosswalk data, and sidewalk conditions point to important next steps for making virtual pre-navigation a practical everyday tool.
Tags: blindness · navigation · cognitive mapping · virtual navigation · orientation and mobility · spatial cognition · assistive technology · wayfinding