Real-time autonomous magnetic microrobot navigation across dynamic and biologically relevant environments

Abstract

Autonomous microrobots could enable minimally invasive interventions in confined biological environments, but their operation requires real-time navigation among moving obstacles and environmental disturbances. Here we present a closed-loop framework for autonomous magnetic microrobot navigation that separates long-range geometric planning from short-range reactive control. An analytic geometry-based planner continuously generates collision-free global routes from microscopy images, while rule-based or reinforcement-learning local controllers respond to moving obstacles before returning control to the replanned global route. The planner produces short paths with substantially lower computation time than representative graph-, sampling- and geometry-based planners, and the integrated framework operates within the video-rate control budget. We demonstrate microrobot autonomous navigation in static and dynamic microfluidic environments, including avoidance of moving particles and human red blood cells, navigation with and against fluid flow, and repeated long-duration target transitions. We further demonstrate magnetic microrobot actuation in ex vivo ocular and vascularized cortical tissues and photoacoustic detection of magnetically induced microrobot dynamics in post-mortem tissue. These results establish a modular framework for autonomous magnetic microrobot navigation in complex biological environments.