
Context: The society and market needs
The zenithal perspective is the only way to monitor exhaustively the whole protection zone of linear infrastructure such as roads (or railways or pipelines,…). The drone point of view allows seeing the roadbed surface, the horizontal traffic signs, the buffering area including invasive vegetation or landslides risky slopes, and traffic flows. This perspective can complement other techniques such as mobile mapping, which is perfect to detect vertical traffic signs, or IoT infrastructure sensors such as car counters.

Objectives
The StratoTrans Project objective is to provide new information to the road and traffic managers, to prevent and to solve both infrastructure and mobility troubles. Many public or private resources are involved in this issue, regarding the conservation or exploitation of the infrastructures, but also the time of the citizens who daily get messed in traffic jams.
It is worth to note that most of the specific requirements come from infrastructure and traffic managers, like public agents Generalitat de Catalunya (Departament de Territori i Sostenibilitat (DTES) – Direcció General d’Infraestructures de Mobilitat (DGIM)), Dirección General de Tráfico (DGT), Servei Català de Transit (SCT), or private agents (GLOBAL-LOCAL). Following the manager’s needs, Exodronics focuses on find the solutions with smart drone technologies.

Methodology
Then the methods of StratoTrans have a double aspect: 1/ Computer Vision (CV), Artificial Intelligence (AI), Remote Sensing (RS), Geographical Information System (GIS): On one hand we systematically acquire imagery with a fixed wing EXO C2-L, following a corridor flight plan over the selected segments of a vector road graph-map, to obtaining updated mapping products (Digital Surface Models, Orthoimagery, thematic maps,…). This big amount of data, characterized by a very high spatial resolution and a specific point of view (different of classical mobile mapping), feeds data pools used train recognition algorithms. The generation of this data pool by itself is of great value for the project, but the core is the use of CV/AI algorithms in the automatic detection of risky slopes to prevent landslides (or rock falls), the aid in vegetation encroachment detection for conservation, or the asphalt maintenance alerts.

2/ 5G, CV, AI, IoT: On the other hand we focus the traffic management. Video imagery is send via 5G from the drone to the traffic control headquarters. Now, for us, the drone is considered as a part of the IoT system, crucial to provide the real-time imagery of a roundabout with the virtue of ubiquity.

Expected results
The goal of this project is to develop a complete solution to monitor the roads as a whole to prevent and solve infrastructure and traffic management problems.
The results will be maps and orthoimagery integrated with official Geographic Information Systems, adapted to the manager’s previous monitoring systems and accessible on-line. Also, we will develop Computer Vision algorithms based in Deep Learning techniques to detect events and quantify traffic flows from video streaming, in order to provide crucial information and help the traffic managers in their decisions.
