Mike has been developing this robot at the CMU NanoRobotics Lab for at least the last 5 or so years. This is the same lab I used to do research for back in my undergrad years.
Wall climbing robots can perform inspection, observation, repair, and act as communications relays. Their small size allows them to enter areas that are inaccessible to humans. The tri-leg Waalbot uses dry adhesion to stick to walls and ceilings as it climbs. The tri-leg design uses simple rotary actuators for a single degree of freedom motion, but includes passive joints and elastic flexures to allow this motion to provide the preload and peeling forces necessary to climb using dry adhesion. A PIC microcontroller is used to control the motion of the robot and onboard power makes the system fully tetherless and wireless RF or IR control allows for teleoperation. The Waalbot design, from Carnegie Mellon University, is simple, meaning it can be easily miniaturized and controlled. The legged motion allows the wall climbing robot to overcome small obstacles and low angle surface curvatures. Even with this low complexity structure, unlike most other climbing robots, the robot is agile and fast with a small turning radius and plane transitions.http://engineeringtv.com/blogs/etv/archive/2008/06/12/waalbot-wall-climbing-robot.aspx