In 2008, the U.S. National Science Foundation funded an agency called the Center for Compact and Efficient Fluid Power that was tasked with developing the next generation of fluid powered devices and technologies. As part of this program, I designed, built, and tested a pneumatic quadrupedal walking robot.
In the NSF's parlance, this robot is a "test bed" meaning that it is intended to demonstrate other technologies developed in the Center in a real-world scenario. This robot fulfills this mission by serving as a platform to test small, mobile, pneumatic power sources. The intention is to prove that fluid powered systems can be devised that are competitive with far more ubiquitous mobile actuation schemes that use electric motors and batteries. This is possible because pneumatic actuators have a much greater power density than electric motors. One way of thinking about this is to remember that electric motors are extremely weighty due to them being filled with copper wiring while pneumatic actuators weigh very little since they are filled only with air.
However, much more work has been done by the global scientific community to develop highly energy dense batteries than has gone into comparable development of mobile pneumatic energy sources. It is these sources that the Center intends to develop and to test on this robot.
The robot has four legs, each with four articulated and actuated joints. Each joint is driven by a bidirectional pneumatic actuator which turns it in a common slider-crank mechanism. Also included in this mechanism is a passive hydraulic damper which moves in parallel with the pneumatic actuator. Although this mechanism degrades the efficiency of the actuation, it is critical because it enables a simple control algorithm for the joint positions. The use of this damper is grounded in efforts undertaken and documented in an ASME Journal of Dynamic Systems, Measurement, and Control article.
The legs are driven in position control modes. They are made to track periodic orbits that are fixed in speed and in mutual phase difference. While many modern robots utilize far more sophisticated modes for controlling their legs, this simplistic method is appropriate for the task (demonstrating mobile fluid power supplies) and has been shown to work very well, when parameters are situationally optimized.
Using the proven control techniques and the inherent advantages of its pneumatic actuators, the robot has been made to perform a variety of locomotion feats. It has achieved both walk (or crawl) and trot gaits, both at a number of speeds. Beyond this, it has demonstrated these gaits while traversing diverse substrates, such as linoleum, concrete, brick, grass, and pine needles.
Demonstrations of its abilities culminated with ascertaining its maximum values of "normalized output power". The use of these metrics allows the easy comparison of walking robots of any size. Using a trot gait, the robot reached a maximum speed of one body length per second. In a different experiment, the robot supported a mock payload weighing 130% of the robot's own weight while walking. These achievements put the robot at the peak of state-of-the-art walking robots, (as of 2010) second only to Boston Dynamics' remarkable BigDog.
