RESUMO: The project will build the first large-scale, multimodal dataset of pedestrian crossing behavior in Rio de Janeiro, captured with reconfigurable environmental intelligence sensors (REIP) that offer 4K video, audio, and traffic signal telemetry. Twenty intersections, purposefully selected for socio-economic diversity and infrastructural contrasts, will be monitored during multiple dayparts and weather events.
An end-to-end, open-source AI pipeline will extract privacy-preserving analytics, including pedestrian trajectory, wait time, gait type, vehicle speed, honk frequency, and signal compliance. Derived indicators will feed exploratory dashboards for urban planners and power proof-of-concept Augmented-Reality (AR) applications on devices like Apple Vision Pro and Meta Quest, demonstrating personalized safety cues and inclusive navigation.
Expected outputs include (i) a publicly released, quality-controlled dataset, (ii) peer-reviewed publications on equity-aware mobility analytics, and (iii) AR prototypes validated with community stakeholders and policymakers. The long-term vision is to inform equitable signal timing, infrastructure retrofits, and responsive traffic-management policies across Brazilian cities.

PESQUISADOR(ES) LÍDER(ES) DO PROJETO*: Jorge Poco
ESCOLA(S): EMAp