Optoelectronic-Guided Automated Navigation of Janus Micromotors in Unstructured Environments

Active micromotors with optoelectronic guidance that are capable of autonomous motion have garnered significant interest, particularly in the field of targeted drug delivery, detoxification, and immune-sensing, etc. However, time-varying uncertain fluctuation in self-propelling velocity can cause active micromotors to encounter unexpected accidents or deviate from their preset state. Here, we propose a novel navigation method for Janus micromotors with guidance of optoelectronic virtual electrodes, involving visual recognition, decision making and motion control. A deep learning model detects the real-time state of Janus micromotors, providing position and velocity feedback for path planning and motion control. Velocity control is achieved by dynamic regulation of the electric peak-to-peak voltages, with tracking errors eliminated through the proxy-based sliding-mode control (PSMC) framework. To avoid obstacle interference, motion strategies for Janus micromotor are formulated by the Hierarchical Value Iteration Networks (HVINs). The Janus micromotor enabled navigating through a confined space containing multiple obstacles and following arbitrary velocity functions with small errors. Simulation and experimental results demonstrated that our navigation method achieves high accuracy in single-micromotor motion control and path planning, which holds significant promising for intricate tasks in biomedical applications.