Now this is why I don't just go for that SF buzzword prefix 'nano'. It's great for 'technology indistinguishable from magic' but with just a little thought you soon realize that microscopic robots should come first and will be very useful. For example, dealing with cancer doesn't have to be a nanoscale operation unless you actually want to repair the DNA faults responsible. Microbots in the body should be able to nip in and zap the relevant cells.
Journal of Applied Physics - Precise manipulation of a microrobot in the pulsatile flow of human blood vessels using magnetic navigation system (3 pages)
There are many other researchers working on bloodstream robots but we are still about 3 to 5 years away from clinical use because of the regulatory timeline.