Lists
Lists
Visha, Bajram, Das, Sonjoy Ranjon, Yasin, Amanullah, Hassan, Bilal and Fernando, Sandra (2025) An AIoT-based real-time bee monitoring and wasp mitigation in apiaries Yolo, Hailo, Honeybee, Sort, Wasp. In: 2025 8th Conference on Cloud and Internet of Things (CIoT), 29-31 October 2025, London, United Kingdom.
Honeybee colonies are increasingly vulnerable to wasp predation, particularly during late summer when resource scarcity intensifies interspecies conflict. Conventional deterrence methods—such as chemical repellents, passive traps, and manual intervention—are inadequate for continuous, selective, and automated protection. This study presents an AIdriven system for real-time detection and deterrence of wasps using embedded edge computing. The system is deployed on a Raspberry Pi 5 integrated with a Hailo-8 AI accelerator (26 TOPS), employing a YOLOv11n object detection model compiled through the Hailo SDK for low-latency inference. Insects are tracked using the SORT (Simple Online and Realtime Tracking) algorithm, which combines a Kalman filter for motion prediction with Hungarian assignment for frame-toframe identity, and targeting is governed by spatial and temporal constraints: a wasp must occupy a predefined lock zone before a Class 3B laser is activated via IRLZ44N MOSFET, with aim managed by dual-axis MG996R servos using a PCA9685 PWM controller. Bees are identified but excluded from all deterrence logic to ensure species' safety. Experimental results demonstrate real-time performance, accurate discrimination, and modular extensibility. The system demonstrates a closed-loop intervention pipeline where realtime perception directly drives physical deterrence with explicit pollinator safety gating-an underexplored integration in apiculture automation.
 |
View Item |