Abstract
In the transition to renewable energies, the presence of large birds and birds of prey leads to conflicts of interest between nature conservation and achieving energy targets, especially in the construction of wind farms. The birds at risk of collision with wind turbines are frequently protected species, which presents a barrier to gaining approval of suitable areas and to the actual siting of new wind farms. The implementation of innovative anti-collision systems, particularly AI-supported radar or stereoscopic camera systems, could reduce the risk of bird collisions with wind turbines. We present an independently controlled empirical study examining the bird detection rate and species classification accuracy of two technically different remote-sensing sensor types: a stereo-camera system and a radar scanning system. The results support the hypothesis that modern anti-collision systems have the potential to detect, recognize, track and correctly classify bird species. They further underline that anti-collision systems can be effective in reducing avian collisions with turbine blades, particularly if the all-weather radar's multi-object, close to 360⁰ scanning capability is combined with the high accuracy of AI-based image recognition from stereo-camera systems.