Robots with sizes comparable to cells will be particularly useful for biological science and medicine. For example, they could simultaneously observe many cells, in vivo over extended periods to time, and the chemical signals they exchange. Such robots could also provide detailed diagnostics and targeted interventions for medicine.
Realizing the potential of these machines requires developing technology to fabricate them in large numbers, cheaply and reliably, and suitable control programs. This talk will focus on control, highlighting how it differs from controlling conventional robots. These differences include physics dominated by surface forces and Brownian motion rather than inertia, the need to coordinate large numbers of robots, and the nature of their operating environments.
Simulation studies incorporating these features identify tradeoffs among task performance and hardware capabilities such as power use and sensor accuracy. Such simulation results can help focus future fabrication engineering on the most promising combinations of capabilities.
I'll illustrate these ideas for a prototypical task: finding the source of a chemical signal introduced into small blood vessels, e.g., due to tissue injury, using typical values for physical properties such as flow speeds and chemical diffusion. The robots may detect signals and initiate response more rapidly than immune cells. They could identify the signal's cause (e.g., a type of infecting bacteria) and, unlike the cells, communicate that information to an attending physician, providing earlier and more accurate diagnosis. This example illustrates the physics involved and quantifies robot performance. More generally, simulating typical task scenarios demonstrates how coordinating microscopic robots differs from approaches used for current, larger-scale, robots.
Schematic showing a one-micron long robot, at upper left, in a small blood vessel with red blood cells.