The Infrastructure branch of the French National Railway, SNCF Réseau, has initiated an innovative project with the Institute Geomatics of FHNW (Basel, Switzerland) to integrate advanced AI and geospatial technologies for predicting wildlife collision risks on its railway tracks. Our initiative responds to increasing wildlife-related disruptions, which cause substantial safety risks and operational delays, by employing a sophisticated data analysis framework that utilizes multimodal dataset classifications.

The focuses lies on modeling wildlife habitats of boar and deer and predicting their movements near railway lines. By analyzing extensive geospatial datasets, including time-series from multispectral satellite remote sensing, railway GIS and on-site camera traps, the project utilizes machine learning algorithms, particularly deep neural networks, to generate dynamic risk maps. These maps indicate high-risk zones where wildlife is likely to intersect with railway operations. This predictive modeling will allow the design of effective mitigation strategies such as wildlife crossings and specialized fencing, directly addressing the dual challenge of ensuring both rail safety and biodiversity conservation. The outcomes aim to significantly reduce the frequency of wildlife-train collisions, enhancing the reliability of rail services and minimizing ecological disruptions.

The project’s methodology stands out by merging rigorous scientific inquiry with practical application. By presenting this work at the IENE 2024 Conference, the research team aims to contribute substantive findings to the discourse on sustainable transport infrastructure that safeguards biodiversity. This approach not only highlights the potential of AI in environmental management but also sets a benchmark for future projects at the intersection of technology and ecology.

FLOX Robotics is a Swedish deep tech, biodiversity startup spinoff from Royal Institute of Technology KTH in Stockholm that transforms human-wildlife conflicts into coexistence, by designing, developing and deploying autonomous cutting-edge technologies for various transportation sectors, including airports, railways and roadways. By harnessing the power of artificial intelligence (AI), robotics and autonomous technology, we develop and deliver effective and innovative mitigation measures. FLOX is named as Top 10 Cleantech company in Europe 2024 by Cleantech Scandinavia and aims at keeping wild animals away from trafficated areas - from reducing bird strikes at airports by using advanced autonomous drones equipped with proprietary wildlife herding technology, proactively deterring wildlife from critical areas such as runways, ensuring airport operations are safer and more sustainable; to stationary systems place alongside railways and roads at wild animal collisions hotspots.
FLOX solutions has proven 100% effective in reducing wildlife collisions in trials conducted at high-risk Swedish airports including Kiruna, Umeå, and Malmö, part of a collaboration with Swedavia that operates 10 major Swedish Airports.
Furthermore, FLOX uses similar technologies to address vehicle collision challenges on roads and railways. Partnering with train manufacturer Alstom, and Swedish Transportation Agency we are now adapting our innovative solutions to create safer travel corridors and prevent wildlife-related incidents on railways. Our integrated approach to wildlife management is also being deployed along roads, where the intersection of animal habitats with traffic poses significant risks.
By leveraging our AI-driven technology, FLOX provides a dynamic and non-invasive method to manage and redirect animal movements, effectively reducing

A new prototype of smart road sign to minimize animal-vehicle collisions (AVC) was installed on 11 road sections in the Region of Castilla y León (Spain). This new generation of vertical signs relies on new technologies to provide real-time information about the likelihood of an animal entering the roadway. To this end, it makes good use of all the scientific knowledge generated in this field and is based on the development of spatiotemporal models that quantify the effect that seasonal and daily patterns, lunar phases, weather conditions, or hunting activity have on the AVC occurrence. An application installed on a server uses these models to calculate the probability of an AVC for each section at any given time and, depending on the established risk thresholds, sends a signal to the LED panels that keep the road users informed. There are three possible levels of alert: off, orange-high risk, red-very high risk. By concentrating warnings only at certain times, the prototype is expected to reduce habituation by drivers, which minimizes the effectiveness of conventional signs. Based on the results provided by a BACI approach, this prototype has reduced AVC by 22.5% on these sections, about 50 fewer AVC than expected, which in economic terms represents a benefit of more than 150,000 euros in avoided material damages. Furthermore, in 2023 there have been no personal injuries on these sections, which did occur in the three years prior to the installation of the signs.

Although electrical infrastructures can be dangerous for avifauna, they are also used as nesting sites by several bird species. This is the case for the White Stork (Ciconia ciconia), whose populations are spatially and numerically expanding in France (population increase +694.6% between 2001 and 2019). This situation affects both the birds and Transmission System Operator (TSO) because it simultaneously puts at risk their populations and electricity quality.
A country-level census led in 2021 and 2022 recorded 5,212 White Stork nests among which 351 were located on electric high tension (HT) voltage pylons. By using robust data, our study aims at quantifying the probability and speed at which White Storks will colonise HT pylons by 2030, using an empirical modelling approach.
To this end, we first identified habitats favoured by storks and compared them with HT pylon locations. Results suggest that total stream length was the main factor affecting the probability of presence of nesting sites. The second and third most important variables were the percentage of urban areas and water surface. However, spatial modelling indicated that White Stork habitat preferences differed between zones across the country: in Alsace, there was a preference for potential foraging territories containing urban areas, contrary to other regions. There was also differential use of HT pylons between zones.
Next steps will explore the drivers explaining why some HT pylons are chosen as nesting sites and establish population distribution scenarios to help the electricity transmission company minimise their impact on birds and guarantee electricity quality.

The nesting of white storks (Ciconia ciconia) on transmission power line pylons is a common and increasing behaviour in many countries. To mitigate outages, Transmission System Operators (TSOs) frequently invest significantly on the installation of deterrence devices at hazardous locations of pylons, but scientific evidence of their effectiveness is still lacking. In this experimental study, framed in a Portuguese partnership between a research centre (BIOPOLIS-CIBIO) and the national TSO (REN), we evaluated the effectiveness of the most commonly used deterrence device by the company (a wind-rotating metallic structure; hereafter “anemometer”) in discouraging storks from building nests.
In a region where REN-pylons hold 1500+ stork nests (near Lisbon, Portugal), 43 nest-pairs existing (on platforms) on 36 pylons were selected. After their removal (in 2020-2022, before each breeding season), an “anemometer” was installed in half of the sites (experimental) and the remaining half kept without deterrent (control), in a paired-sampling at the pylon-scale. All sites were then regularly monitored throughout the following two breeding seasons.
The overall nest reconstruction rate after the first season was significantly lower in experimental (42%) than in control (74%) sites, a pattern that kept similar in the second season. Focusing on the subset of (18) pairs most representative of real locations where “anemometers” are installed on pylons, an effectiveness of 47% was obtained. This study also showed that this value (although not being matter of concern, given the high success of overall REN outage-preventing nest-management program) can be improved by reducing the lateral spaces between adjacent “anemometers”.

Collisions of birds with transparent and reflective materials are one of the most common causes of death of birds of anthropogenic origin. Significant collision areas also include noise barriers along line structures.
The presentation captures the development of solutions to the issue. The first local studies (Moravian-Silesian, Zlín Regions) carried out by the "before-after monitoring" method showed a more than 90% reduction in mortality compared to the baseline. Subsequently, general recommendations for adequate security were formulated, consisting in the selection of suitable materials for newly constructed barriers (screens), or parameters for additional security of existing noise barriers. These recommendations were subsequently implemented into the Technical Condition No. 104 of the Ministry of Transport of the Czech Republic, which entered into force in 2016.

Road infrastructure and related traffic significantly affect the various habitats of bird assemblages in their vicinity, all over the world. We examined the impact of highways on the qualitative and quantitative characteristics of bird assemblages in agricultural landscape with scattered tree and shrub vegetation. The number of bird species was recorded at 298 observation points which involved four different distances approximately 25, 125, 500, and 1000 m far from the edge of highway. We found out that the bird species diversity decreased towards to highways. This decreasing was more significant around highways with higher traffic volume. It is assumed that the main cause of the decrease in bird diversity is noise, which is correlated with highway distance and traffic volume, partly because farmland can facilitate the propagation of noise. Results also show that farmland birds were influenced by the highway distance slightly more compared to woodland birds. Further we observed the strongest impact of the highway on shrub-nesting species, while other nesters were less affected. We also found high significance of negative effect of highway distance on insectivorous species. For granivorous species, a detrimental effect was also noted, however it was less pronounced than for insectivorous species.