A French research team has developed a technique to manage collisions between animals and vehicles. The technique takes advantage of the opportunities presented by the growing number of sensors integrated into transportation infrastructures and the creation of their digital twins. Using a network of camera traps, the objective is to map the collision risk between trains and ungulates, such as wild boar and roe deer. The study was published in the open-access journal Nature Conservation.
Example of a map showing the estimated relative abundance of a species along a railway section. The higher the abundance, the higher the collision risk. Image Credit: TerrOïko
Animal-vehicle collisions pose a risk to human safety and conservation efforts, and they are extremely expensive for both transportation infrastructure users and managers.
Sylvain Moulherat and Léa Pautrel led the study from OïkoLab and TerrOïko, France.
The suggested approach starts by using ecological modeling software to simulate the most likely animal movements inside and outside an infrastructure. This allows us to determine where they will most likely cross.
After identifying these collision hotspots, ecological modeling is once more employed to support the field deployment design of photo sensors. A number of scenarios are modeled to determine which deployment scenario's expected outcomes are most in line with the initial simulation.
Following sensor deployment, artificial intelligence (deep learning) processes the data (in this case, photos) to identify and detect species in the area around the infrastructure.
Finally, the processed data are input into an abundance model, a different ecological model. It estimates the likely density of animals in different areas of a studied region based on data collected from only a few points within that area. The outcome is a map depicting the relative abundance of species, which helps assess the collision risk along an infrastructure.
Although this approach was used on a real railway segment in southwest France, it can be used on any transportation infrastructure. It can be applied to both new infrastructure development and existing infrastructure (as part of the environmental impact assessment strategy).
This approach opens the door for incorporating monitoring systems focused on biodiversity into transportation infrastructures and their digital twins. In the future, sensors that continuously gather data may be enhanced to generate dynamic adaptive maps and real-time driver information, which could eventually be transmitted to autonomous cars.
Journal Reference:
Moulherat, S., et al. (2024) Biodiversity monitoring with intelligent sensors: An integrated pipeline for mitigating animal-vehicle collisions. Nature Conservation. doi.org/10.3897/natureconservation.57.108950