Résumé
How we walk at home, in a restricted and cluttered environment, reveals crucial information about locomotor efficiency and wellness. Gait trajectory, walking metrics and space occupancy are for instance useful markers of our daily locomotor routines which are informative about well-being. The goal of this study was to detect, reconstruct and identify locomotor trajectories during walking at home with a capacitive sensing system, the SensFloor. The study was conducted in an apartment located in the centre of Montpellier in the south of France. Data were first collected in controlled linear and curvilinear walking scenarios performed by three participants. These data allowed the construction of the processing algorithm through the following the steps: (i) defining the static graph of the apartment, (ii) identifying spatio-temporal neighbourhoods from ground activation, (iii) data filtering, (iv) trajectory reconstruction from the chains of dynamic nodes allowing finally (v) identification of walking metrics: locomotor distance, time and speed. In a second phase, our algorithm was tested on a full day of real occupancy by one single inhabitant. The results demonstrated the effectiveness of our algorithm in identifying locomotor trajectories within the apartment. A random one-day file extracted from the database revealed 529 individual trajectories, computed from 100,490 floor activation contacts. Trajectories were produced with an average speed μ_v=1.4 m/s, for an average travelled distance (per trajectory) of μ_l=14.5 m and an average walking time of μ_d=9.4 s. Each identified trajectory had an average of 50 activations. In addition, the extracted temporal information showed a dominant occupancy of the living room and room 1, for a total travel time of 8.21%. The success of our algorithm in detecting trajectories and identifying metrics of interest now allows us to explore locomotor walking habits in reallife conditions, and in the face of current problems such as home confinement related to Covid-19. Acknowledgments: This research was conducted as part of the Human at home projecT (HUT) co-financed by the European Regional Development Fund (ERDF) and the Occitanie Region. This acknowledgment extends to the support given by Montpellier Méditerranée Métropole.