Estimating geomagnetic field detection sensitivity of pigeons and passerine migrants using deep machine learning
- P 388-395
The capability of homing pigeons and passerine migrants to derive navigation-rated information from the geomagnetic field (GMF), enabling them to navigate from unfamiliar sites, is a subject of research. Despite the vast trajectory information available from field experiments, the true accuracy of their map and compass sense are seldom reported. The recent developments in bird geo-tagging, precision world magnetic model and deep machine-learning capabilities enable us to understand the mechanisms underlying their innate abilities in true navigation. In this article, we machine-learnt the GMF anomaly in a 10 km2 region and analysed the flight path efficiency for a 10-km GMF-anomaly-guided trajectory using the developed deep-learning-based artificial intelligence (AI) algorithm. From the simulated flight path and comparing the computed efficiencies with the field-reported results, it is observed that the sensitivity of the GMF gradient detection sensory system in pigeons and passerine migrants in familiar and unfamiliar regions are in the range of 1–3 nT and 0.5–2.5 nT respectively. Identified results shall help implement AIbased solutions for understanding spatiotemporal bird migration and enacting environmental conservation policies.
Deep learning machine learning magnetic field navigation