Large-Workspace and High-Resolution Magnetic Microrobot Navigation Using Global-Local Path Planning and Eye-in-Hand Visual Servoing

Abstract

Magnetic microrobots have the capability of navigation in confined and narrow space to perform delivery or micromanipulation tasks. However, due to the fast decay of magnetic fields and the contradiction between resolution and field of view (FOV) of a feedback instrument, the large-workspace and high-resolution (LWHR) navigation remains a challenge in magnetic microrobotics. This paper provides a solution to this challenging problem, in which a self-constructed magnetic manipulation system with mobile electromagnetic coils and eye-in-hand feedback is used to actuate magnetic microrobots in a large workspace. To enable the automated large-workspace microrobot navigation in complex environments, we propose a global-local path planning scheme. In the mode of low-resolution feedback, the whole workspace is captured and a global near-optimal path is planned. On the other hand, in the high-resolution feedback mode, the microrobot and its local environment is precisely tracked and identified, respectively. Then, a real-time local planning algorithm is designed to correct the deficiencies in the rough global path. Parameter tuning of the planning scheme is accomplished via simulations. Closed-loop motion control and field control algorithms are designed and implemented, and experiments demonstrate the automated LWHR navigation of magnetic microrobots in a vascular-like network. Results show that the ratio between the navigation diameter and the microrobot diameter exceeds 200.