HuBot: A biomimicking mobile robot for non-disruptive bird behavior study

Abstract

The Houbara bustard, an avian species of conservation concern, poses significant challenges to researchers because of its elusive nature and sensitivity to human disturbances. Traditional research methods, often reliant on human observations, face some challenges and can inadvertently affect bird behavior. To overcome these limitations, we propose the HuBot, a biomimetic mobile robot designed to seamlessly integrate into the natural habitat of Houbara. By employing advanced real-time deep-learning algorithms, including YOLOv9 for detection, MobileSAM for segmentation, and vision transformer (ViT) for depth estimation, HuBot semi-autonomously tracks individual birds, providing unprecedented insights into their individual behavior, social interactions, and habitat use. HuBot can thus contribute to a deeper understanding of Houbara behavior and its ecology. The biomimetic design of the robot, including its life-like appearance and movement capabilities, minimizes disturbance, allowing for monitoring of Houbara birds while minimizing disruption to their behavior. Rigorous testing, including extensive laboratory experiments and field trials on challenging terrains, validated the performance of HuBot as a complementary tool for traditional observation methods. • Mimics Houbara bustards to reduce disturbances and enable natural monitoring. • Uses YOLOv9 and MobileSAM for real-time tracking and habitat analysis. • Captures behaviors in real-time, eliminating the need for post-processing. • Integrates ecological data for accurate and adaptive observation models.