In context
In late 2023, magnetic microrobots were emerging as a promising tool for minimally invasive medical procedures, but precise trajectory tracking in complex fluid environments remained a key challenge. Traditional model-based controllers struggled with uncertain dynamics and external disturbances, limiting practical deployment.
What was reported
Researchers at Jiangnan University proposed an adaptive neural network control strategy for magnetic microrobots driven by external magnetic fields. The method uses a radial basis function neural network (RBFNN) to estimate system uncertainties online, eliminating the need for precise dynamic models. Stability was proven via Lyapunov theory, ensuring bounded tracking errors.
Experiments were conducted on a custom electromagnetic system with Helmholtz and Maxwell coils, using a cylindrical NdFeB microrobot (2 mm diameter, 3 mm height). The controller achieved root mean square errors of 6.2204 pixels for curved trajectories and 6.4279 pixels for straight lines, significantly outperforming a traditional PID controller.
The neural network weights adapted quickly, allowing the microrobot to converge to desired paths even with unknown fluid resistance and dynamic parameters. The system operated at 30 Hz control frequency, with real-time Python-based control.
Why it mattered
This work demonstrated that neural adaptive control can handle nonlinear, uncertain dynamics in magnetic microrobots, offering a robust alternative to model-based methods. It paves the way for more reliable autonomous operation in medical applications such as targeted drug delivery and microsurgery, where precision is critical.
“The results demonstrate that the root mean square errors for trajectory tracking in curved and straight paths achieve 6.2204 pixels and 6.4279 pixels, respectively, significantly surpassing the performance of the traditional PID algorithm.”
Source: 《机器人》期刊 (robot.sia.cn) · Published 2023-11-10 · “磁性微型机器人神经自适应轨迹跟踪控制”