Path Planning and Tracking for an Underactuated Two-Microrobot System

We propose a robotics method to independently control the position of multiple magnetic microrobots (agents) through their local magnetic interactions in close proximity. Utilizing a rapidly-exploring random tree (RRT) planner, we generate a trajectory which is then followed by an optimization-based controller (OBC), while avoiding obstacles. The method generates a path that obeys the governing unique nonholonomic actuation constraints in the agents' motions, and has potential to scale to more agents. This is the first work of its kind to introduce a method for autonomous navigation of multiple magnetic microrobots in close proximity with obstacle avoidance. The ability to independently control the motion of multiple magnetic microrobots in this manner makes the method useful for potential applications such as targeted drug delivery and medical diagnostics. We prove that the two-microrobot system studied here is small-time locally controllable (STLC). We conduct two practical demonstrations to verify the ability of the RRT-based navigation applied to a pair of magnetic microrobots moving in a microchannel: 1) multi-agent micro-factory, and 2) team-targeted cargo delivery. An average tracking error of around one-fifth of the agents' body-length is achieved for the control of agents' positions.