InMob Dataset: Walking data from visually impaired and normally sighted pedestrians

The dataset was collected through the ANR project, Labcom InMob, that focused on developing positioning and navigation algorithms specifically tailored to people with visual impairments (PVI) by integrating personalized mobility profiles into their design. To achieve this, we needed a large quantity of walking data from both normal sighted people (NSP) and visually impaired volunteers, which is used to create mobility models using artificial intelligence (AI). To assess the stability of the mobility models over time, the same volunteers repeated the experiments over months or years. The dataset, collected between October 2021 and May 2025, covers approximately 450 km of walking data, of which about 150 km is collected from PVI. We collected the data with ULISS, our self-developed navigation system. As the research progressed, we started using smartphones for data collection. The collected data were used to develop and evaluate a personalized PDR framework MAPIN (Fu, et al 2024), to position people in city conditions (Wang et al 2024), to detect crosswalks through inertial data (Danis, et al 2024) and to correct PDR framework using detected crosswalk information (Danis, et al 2025) so far.