gaips_bea image1 image2 image3 image4 image5 gaips_ecute_beach_bar_banner gaips_ecute_train_incorrect_ticket_banner
iLU - Integrative Learning from Urban Data and Situational Context for City Mobility Optimization Webpage icon

Collaborators: Francisco S. Melo, Alberto Sardinha, Guilherme Varela


Mobility of passengers and freights in most European capital cities such as Lisbon is not yet sustainable. Road traffic is associated with significant externalities (social, economic and environmental costs) due to daily congestion, traffic noise, air pollution, accessibility problems, and road accidents. In addition, mobility data is dispersed through several entities/operators. The iLU project aims to (i) consolidate the multiplicity of data sources on city mobility stored in the Plataforma de Gestão Inteligente de Lisboa (PGIL) and guarantee its real-time updatability; (ii) discover actionable spatiotemporal patterns of mobility from such heterogeneous data sources, particularly non-trivial correlations between road traffic and situational context data; (iii) anticipate traffic congestion using advanced and integrative predictive models; (iv) real-time support of mobility decisions through the use of deep reinforcement learning to positively condition the city traffic by, for instance, controlling traffic lights and road message panels.


A Methodology for the Development of RL-Based Adaptive Traffic Signal Controllers
bib icon
Guilherme Varela and Pedro Pinto Santos and Alberto Sardinha and Francisco S. Melo, AAAI Workshop on AI for Urban Mobility (AI4UM), February, 2021