I know the line is blurry, but I disagree that clumps of iron oxide nanoparticles rise to the level of "robot"
If there's offboard sensing and control systems which can maneuver the particles individually then you could make a case for it. If you're just jiggling them with a magnetic field then no.
For example, the best paper award at the Robotics: Science and Systems conference this year was won by a magnetic jellyfish: http://www.roboticsproceedings.org/rss15/p13.pdf
Maybe the magnet should be counted as part of the robot, with the moving parts being analogous to a gripper.