Traffic Intelligent Drone Evaluation using Artificial Intelligence

The solution we propose in this project responds to the challenge set by Fingal County Council in Ireland to evaluate a “quiet street”. It is one of the most effective and efficient solutions that have been developed so far. The project carries out a simultaneous observation of all the required metrics, including traffic volume, patterns, movements, speeds, behavioural patterns, interactions between road users, modal share and near conflict detections. This is done with a single observation technique developed by integrating drone and artificial intelligence techniques. Air quality data will also be recorded. The technology can be easily replicated and is therefore ideal for conducting multiple effect evaluations (before, during and after intervention).
The project will provide detailed data on the following parameters:
• Traffic volume – accurate classification and counting of road users including pedestrians, bicycles, motors, cars, vans, buses and trucks. Both raw data and aggregations will be provided.
• Modal share data – correction algorithms are applied on the raw count data for accurate determination of modal share (see section state-of-the-art).
• Movement dynamics – detailed trajectories of all road users will be reported. Origin/destination matrices, near conflict events (using parameters inferred from scientific literature) and visual illustrations will also be reported. All relevant movement dynamics (e.g. near conflicts and behavioural changes) will be automatically and separately saved for post-analysis and further visual inspection.
• Speed data – detailed speed data (average speed, max. speed, min. speed, V85 speed, etc.) will be analysed. It will also be possible to analyse and compare different road segments.
• Various events and time present – detection and reporting of illegal/hazardous parking and non-compliance with traffic regulations, along with detailed reporting on user presence within the observation area.
• Air quality data – we will use street-level sensors to avoid the “diluted emissions” problem that drone technology may experience at higher altitudes.
• Video data.