TECH-4-BIRDS: two AI bird detection systems installed at a wind farm to reduce collisions

Location: “Casa Vieja” wind farm, Palencia

Dates: 2025

Area: Environmental Assessment and Biodiversity Area

In TAXUS MEDIO AMBIENTE we started the year with an important milestone: the installation of two bird detection equipment “TECH-4-BIRDS” in the wind farm “Casa Vieja” in Palencia.

This step marks the transition of the project from the design and laboratory experimentation phases to a key stage: field testing, where the system is put to work in real conditions to further refine its performance and robustness.

“TECH-4-BIRDS is a R+D+i project project promoted by TAXUS MEDIO AMBIENTE together with PXRwith the collaboration of ERBI as owner of the “Casa Vieja” wind farm, and with financing from SEKUENS (Science, Business Competitiveness and Innovation Agency of Asturias).

TECH-4-BIRDS” bird detection equipment installed on the wind farm tower.

TAXUS MEDIO AMBIENTE has more than 20 years of experience in environmental assessment, monitoring and management of wind farms. This project was born with a clear objective: to advance towards technological solutions to reduce bird mortality, one of the most relevant environmental impacts associated with this type of infrastructure.

What is TECH-4-BIRDS?

TECH-4-BIRDS is a bird detection system that combines cameras and artificial intelligence to:

  • Automatic bird identification
  • Analyze your flight path
  • And allow the activation of measures accordingly to minimize the risk of collision.

In addition, the project has two technical goals that are particularly valuable for environmental management:

  • Record the 3D/4D flight path of each bird within the wind farm space.
  • Calculate the time in risk zone, broken down by wind turbine, for each individual entering the monitoring area.
Bird detection by the motorized telescopic camera of the PTZ dome.

How the system is built: “coarse-scale” and “fine-scale” detection

Each “TECH-4-BIRDS” unit consists of two sets of cameras: several wide-angle cameras that cover a wide field of view although with lower resolution in the first phase of bird detection; and on top of that a PTZ dome with a motorized telescopic camera that moves to obtain a high resolution high magnification of the area where the wide-angle cameras detected possible birds in order to confirm the detection and identify the birds.

In this way, two types of cameras are combined:

1) Wide-angle cameras (VR180)

A set of virtual reality cameras with two “fisheye” lenses (200º field of view) for general detection in the environment. They serve to “see” the movement of birds in a wide area and trigger tracking.

2) PTZ dome with motorized telescopic camera

On top of the system a PTZ (pan-tilt-zoom) dome with a telescopic camera is installed telescopic camera camera for fine scale tracking, with the ability to classify and track the bird in more detail.

The equipment is located on towers specifically installed in the park, 12 meters high, to maximize the field of vision and improve detection.

Placement of the “TECH-4-BIRDS” bird detection equipment on the wind farm tower.

AI “on the edge”: acting in less than a minute

A critical part of the design is that the cameras have computational power for local processing of AI models. Why? Because in this application time counts: there may be less than a minute available to trigger a protection measure. Processing “locally” reduces latencies and enables faster response.

Detection areas of the two installed equipment (in blue the overlapping area where it is possible to detect and position birds in 3D with the wide-angle cameras, in yellow the area where it is possible to detect moving birds without positioning or classification with the wide-angle cameras, in red the area where it is possible to follow and classify birds with the motorized telescopic camera of the PTZ dome).

From video to 3D position: triangulation and photogrammetry

To estimate the position of the birds at each instant, the system combines information from several cameras. The idea is:

  • Generate “rays” (lines of sight) from each detection in each video
  • Search for crosses between these rays
  • And thus estimate the 3D position of the bird at each moment (triangulation).

In addition, augmented reality software is used to perform a photogrammetric scan, which helps to spatially contextualize the environment and improve the reconstruction.

Triangulation using rays to estimate the 3D position of birds.

What’s next?

The installation of the two pieces of equipment at the end of 2025 will allow the system to be continuously tested in real-life situations throughout 2026. The aim is to gather evidence of performance, identify opportunities for improvement and evolve the design into an even more reliable and effective product to help reduce collisions in wind farms. reduce collisions in wind farms.

Scroll to Top