So he uses cell cultures as controllers for robots. I don't have access to his article now to see what he did exactly, but I doubt he can make this work efficiently.
I remember a talk about some DARPA projects after 9/11 for chemical sensing that used also alive cell cultures. Advanced chemical sensing could serve to detect explosives and chemical weapons and therefore shield again terrorist attacks. They would take a cell culture from a rat brain put it onto a dish and then characterize the responses to different kinds of currents (output from chemical sensors), so they could implement a computational model that separated these responses and made sense of them. The problem they had was that the cultures are difficult to maintain in the same state. For chemical sensing they had to be transported which was problematic, because even slight jerks would change the culture. They are very sensitive to temperature and then of course they change over time (as the introduction states). So, the would have to be adjusted again, which is time consuming and expensive. Until he finds a way to maintain the state of the network and control the intrinsic changes of the network (culture) he will have problems with stability and it will not be very useful.
With your bare hands?!?