Related Experiment Video
Updated: Jun 22, 2025
14:42
Remote Magnetic Actuation of Micrometric Probes for in situ 3D Mapping of Bacterial Biofilm Physical Properties
Published on: May 2, 2014
9.2K
Magnetic-Controlled Microrobot: Real-Time Detection and Tracking through Deep Learning Approaches.
Hao Li1, Xin Yi2, Zhaopeng Zhang2
1Department of Mechatronics and Information Engineering, Shandong University at Weihai, Weihai 264209, China.
Micromachines
|June 27, 2024
Summary
This study introduces an enhanced YOLOv5 system for precise microrobot detection and tracking, crucial for biomedical applications like targeted drug delivery. The system significantly improves accuracy in identifying small magnetic targets and microrobots in real-time.
Area of Science:
- Robotics
- Biomedical Engineering
- Computer Vision
Background:
- Microrobots offer significant potential in biomedicine for diagnosis, drug delivery, and minimally invasive treatments.
- Accurate detection and tracking of microrobots are essential for their effective application in vivo.
- Existing methods may face challenges in precisely identifying and monitoring small-scale robotic devices.
Purpose of the Study:
- To develop and evaluate an enhanced microrobot detection and tracking system (MDTS) using an improved YOLOv5 network.
- To increase the precision and real-time capabilities for detecting and tracking small magnetic bodies and microrobots.
- To validate the system's robustness and effectiveness for in vivo microrobot operations.
Main Methods:
- Utilized an enhanced YOLOv5 (You Only Look Once version 5) network structure for pretraining.
- Employed magnetic bodies (3 mm, 1 mm) and a 2 mm magnetic microrobot as training targets.
- Integrated a visual tracking algorithm with the improved YOLOv5 model for real-time position and motion information acquisition.
Main Results:
- Achieved 95.81% accuracy for 3 mm magnetic bodies (2.1% increase).
- Reached 91.03% accuracy for 1 mm magnetic bodies (1.33% increase).
- Attained 91.7% accuracy for 2 mm microrobots (1.5% increase).
Conclusions:
- The enhanced YOLOv5-based MDTS effectively enables real-time detection and tracking of magnetically controlled microrobots.
- The system demonstrates robustness and effectiveness in both 2D and 3D environments.
- This technology provides crucial support for the operational control of microrobots within biological systems.
