The Biped
The biped was built on a small undergraduate research grant. The initial goal was to make a biped that could walk dynamically, much like a human walks, rather than statically, like a frame walker. After a number of mechanical modifications and considerable software testing, our first success was to make the biped balance. It stood on a surface (a notebook) that could be tilted, and the robot would stand upright regardless of the angle of the surface. We later succeeded in having the biped stand on one foot, and then shuffle forward. However, we have determined that further software efforts will be relatively fruitless, due to the lack of strength of the servos. Extensive mechanical modifications would most likely be more effort than they are worth; a completely new mechanical platform would be more beneficial. We have compiled all the paperwork into one notebook, and left the biped complete.
This is a rendering of our biped, done by Charlie Reverte. The software control for this robot was Rich Labarca's senior design project for Computer Science. Peter Sand worked on the electronics and software. Jonathan Hurst designed and built the mechanical parts.
This is a picture of the biped, partially assembled. This picture shows the frame and legs from the knees up. The lower legs will be about 1.5 times longer than the upper legs shown. At this point in construction, there was a large bicycle helmet that fitted over the top of the frame. You can clearly see the knee joints on the biped itself, and the ankle next to it. Most of the parts for the ankle are machined, but it isn't yet assembled.
The ankles have two degrees of freedom, radial on the X and Y axes, due to a small differential. The gears next to the ankle will be in the center of the ankle, and act as the differential. The two pieces of metal straddling the square will be attached to one servo each. If they move together, the differential will be locked, and the ankle will move in that direction. If the servos move in opposite directions, the differential will engage, and the gears perpendicular to the original ones will rotate. This gives the ankles two degrees of freedom, even though the servos are mounted parallel to each other. It also gives double the strength in any one direction, because both servos are used for a single motion.
This is a picture of the biped, mechanically finished. Developing the software and walking algorithms will be at least as much work as actually building the robot.
Here's a picture of the biped standing on one foot, in the final stages of mechanical modification. The biped is unsupported, it just looks like Peter's hand is touching it. To walk (or shuffle), it was determined that the ankles we unnecessary, and added too much weight. We removed them, but have since replaced them for storage.
The biped was funded by two Student Undergraduate Research Grants from CMU. The second grant was from Motorola. The project was presented at "The Meeting of the Minds."
If you have questions, please Email me at jhurst@andrew.cmu.edu
back to projects