Development of an automatic beehive transporting system based on YOLO and DeepSORT algorithms

• Published in: Computers and electronics in agriculture Vol. 229; p. 109749
• Main Authors: Wang, Pingan, Kim, Subae, Han, Xiongzhe
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.02.2025
• Subjects: agriculture; algorithms; Automatic beehive transportation system; automation; beehives; Beekeeping; DeepSORT; electronics; Multidimensional vibration analysis; robots; transportation; vibration; YOLO; YOLO; DeepSORT; Beekeeping; Automatic beehive transportation system; Multidimensional vibration analysis
• ISSN: 0168-1699
• DOI: 10.1016/j.compag.2024.109749

•Developed an automatic beehive transport system within a desktop-level robot arm simulation environment.•Created a kinematic model of the robotic arm, along with models for beehive recognition and tracking, complemented by an independent internal and external control algorithm.•Designed a transporting algorithm utilizing PID and PID + MPC, with performance comparisons conducted across various metrics, including runtime and trajectory error.•Achieved impressive results with high accuracy (RMSE error of 11 mm) and efficiency (grasping time of 38 s) in beehive handling. Beekeeping plays a crucial role in global agriculture, yet the lack of automation and an aging demographic require beekeepers to inspect and manually transport beehives frequently. This low level of automation often leads to financial losses and safety risks. To address the challenges of beehive transportation in apiaries, this study developed an automatic beehive transportation system using dual cameras (stereo and RGB) and a desktop-grade robotic arm. The dual cameras served as an information collection unit to initially locate a target beehive, the You Only Look Once (YOLO) and Deep Simple Online and Realtime Tracking (DeepSORT) algorithms were used to identify the beehive, and then the robotic arm was activated to transport the beehive. The performance comparison between standalone Proportional-Integral-Derivative (PID) control and a hybrid PID with Model Predictive Control (MPC) algorithms for a robotic arm was conducted through a detailed evaluation of trajectory error, operational time, and vibration metrics. The analysis aimed to assess the control accuracy, efficiency, and stability of the system under both control strategies. The experimental results found that the system identified beehives with an accuracy of 93.9 %. The combined PID + MPC control algorithm demonstrated superior performance, reducing the trajectory error to 11.05 mm, compared to 11.61 mm with standalone PID control. The control time for PID + MPC was 38.55 s, representing a 16.82 % improvement over PID control, which took 46.35 s. Additionally, multidimensional vibration analysis indicated that PID + MPC effectively reduced vibrations during transportation, enhancing the stability of the beehive-handling process. This study thus demonstrates the first fully automated system for beehive transportation and has great potential for improving automation in apiaries.