Recently, advances in wearable inertial technologies have enabled real-world digital mobility assessment. However, mobility measures are often aggregated across contexts, potentially masking environmental adaptations and age-related differences that may emerge during complex activities and concurrent tasks which require continuous perceptual processing, visual exploration, and motor control. To overcome these limitations, head-worn devices integrating inertial sensors and scene-recording cameras can be used to provide simultaneous information on movement and environmental context, enabling locomotor characterization. This thesis aimed to investigate context- and age-related differences in gait and head-movement characteristics during urban locomotion by developing a pipeline for context identification using egocentric videos acquired through smart glasses.