Multi-modal data usage for smart city mobility services

Abstract

With the rapid increase of population in cities along with connected technologies, cities are made smart to facilitate the populace at their best. Mobility and transport facilitation is the key to a healthy and commutable smart city. With this vision Intelligent Transportation Systems (ITS) have historically been introduced to increase the transportation network performances

allowing for the optimization of several indicators that are strictly related such as travel time, emissions, usability, and safety. The effectiveness of all ITS proposed strategies is mainly based on the ideas of traffic parameter predictions and controlling or anticipating driver’s / traveler’s behavior. Indeed, all relevant policies such as driving guidance, mobility information systems design, and traffic management are based on the consistency between the decision/control variables and the actual traffic parameters (congestion, travel times, vehicle tracking, etc). So, in this dissertation, the goal is to generate intelligence out of structured data for facilitating respective smart mobility services. In the first use case, we leveraged crowdsourced data by road users to get congestion intelligence. Using that, a simple control algorithm for adaptive signaling in developing countries was proposed. Further, we explored socially acceptable route planning for personal mobility devices and safe route planning for pedestrian micro-mobility during the pandemic. For the smart surveillance services use case, we collaborated on robust and domain-invariant online object detection and tracking. In yet another mobility service, aimed at driver assistance using accident event data and corresponding street view images, context-specific accident-prone features are identified. The detected accident-prone features are to be notified to drivers in a proposed head-up display to enhance their decision-making. The ultimate goal is to power new fine-processed data that can be used and reused across applications and businesses, with each data modality adding valuable intelligence one over the other towards autonomous and smart city mobility.