This paper introduces a two-dimensional autonomous navigation strategy of a 750 µm steel microrobot along a complex fluidic vascular network inside the bore of a clinical 3.0 T magnetic resonance imaging (MRI) scanner. To ensure successful magnetic resonance navigation (MRN) of a micro-robot along consecutive channels, the design of autonoumous navigation strategy is needed taking into account the major MRI technological constraints and physiological perturbations, e.g. non-negligible pulsatile flow, limitations on the magnetic gradient amplitude, MRI overheating, susceptibility artifacts uncertainties , and so on. An optimal navigation planning framework based on Pareto optimality is proposed in order to deal with this multiple-objective problem. Based on these optimal conditions, a dedicated control architecture has been implemented in an interventional medical platform for real-time propulsion, control and imaging experiments. The reported experiments suggest that the likelihood of controlling autonomously untethered 750 µm magnetic microrobots is rendered possible in a complex two-dimensional centimeter-sized vascular phantom. The magnetic microrobot traveled intricate paths at a mean velocity of about 4 mm s −1 with average tracking errors below 800 µm with limited magnetic gradients ±15 mT m −1 compatible with clinical MRI scanners. The experiments demonstrate that it is effectively possible to autonomously guide a magnetic microrobot using a conventional MRI scanner with only a software upgrade.