Academia
鈥�
今日黑料 smart-traffic research presented at Transport Research Arena 2026
Dr Liam Butler's presentation, "Optimal Origin-Destination Matrix Estimation from Sparse, Missing and Noisy Observations in Dense Urban Environments", tackled a problem at the heart of modern transport planning, understanding how vehicles move through a road network when the available data is incomplete and imperfect.
This presented work is part of the RoadEye Project, with Dr Kenneth Scerri from the Faculty of Engineering and Dr Odette Lewis from the Faculty of the Built Environment as Principal Investigators, with as partners and in collaboration with Infrastructure Malta.
RoadEye is supported by the Ministry for Education, Sport, Youth, Research and Innovation through the Post-Doctoral Fellowship Scheme 2024.
Why traffic data is harder than it looks
Origin-destination (OD) matrices describe how traffic flows between locations, and they underpin everything from traffic-signal optimisation and infrastructure design to demand and emissions modelling.
Traditionally, building them has relied on costly surveys, fixed sensors or GPS tracking. While traffic cameras are now widespread, in practice their coverage is patchy. Footage is often affected by occlusion, lighting and system downtime, observations are often limited to entry and exit counts, and full vehicle trajectories are rarely available.
Most existing methods assume clean, complete data and degrade quickly when it is missing or noisy. These are the conditions found on real urban roads.
A Maltese collaboration
The study draws heavily on the expertise of Greenroads, a Malta-based company specialising in AI-powered traffic video analytics for smart cities. Greenroads developed the computer-vision pipeline at the core of the research, using a tracking-by-detection approach to identify vehicles, follow them across overlapping camera views and extract counts and trajectories in real time.
This transforms the raw junction footage into the structured data the estimation framework depends on. The company also led the coordination and management of the dataset, a contribution that was essential to evaluating the method under genuine operating conditions.
The traffic data underpinning the study was provided by Infrastructure Malta, the national agency responsible for developing, maintaining and upgrading the country's roads and public infrastructure. Its support gave the team access to a real, busy junction on the live road network, ensuring the framework was tested against genuine Maltese traffic conditions rather than a controlled laboratory setting.
Built in Malta, with sustainability in mind
Beyond the technical contribution, the research speaks directly to sustainability goals. Accurate, real-time OD estimation helps cities fine-tune signal timing, ease congestion and model vehicle emissions more precisely.
These are all essential for decarbonising urban mobility. Since it works reliably from low-cost camera data, the approach is especially well suited to small, constrained networks such as Malta's, where every junction counts and large-scale sensing infrastructure is not always feasible.
It offers a practical route towards the kind of data-driven, climate-conscious transport planning that European research is increasingly prioritising.
