Magnetic Navigation of Artificial Bacteria Flagella in Blood and Water

Artificial bacteria flagella (ABFs) are magnetic helical micro-robots that can be remotely controlled via a uniform, rotating magnetic field. While the control of a single swimmer with the uniform magnetic field is straightforward, the independent control of multiple ABFs is an unsolved problem. We demonstrate such independent control by exploiting the distinct geometries of the ABFs through an analytical solution and a reinforcement learning algorithm. We compare the two algorithms and demonstrate the superiority of reinforcement learning in achieving multiple objectives. The results demonstrate, effective independent navigation of realistic micro-swimmers in a viscous flow field. Moreover we demonstrate the navigation of ABFs in microscale simulations of blood flows.