My Story: Building a Bipedal Walking RobotBy Mark Miller
I am 56 years old and have been building robots and machines since I was eight. Much to the dismay of my parents I became a pack rat and dragged home all sorts of junk to build things with. By 12 I had built my first working TV set using an oscilloscope CRT which went on to win first prize in a science fair.
Back in the 90s I got the crazy idea to build a robot that walked like a person. Up to that point, all of my robots rolled, crawled or ran on tracks just like everyone else's. It wasn't so much I wanted to do something different, but I wanted to mimic human type transport. I had visions of a robot that not only walked on two legs but also one that could take on any number of external features. Different bodies, different heads, arms and functions. I wanted something that would be stable and could have an exoskeleton of my choice. I wanted to build an army of them to have a fleet of magnificent machines.
My first attempts were disastrous. I used joint motors for hips, knees and ankles. Getting them to all sync up was sketchy at best. The only thing my robot knew how to do was "fall over." I tried lots of stuff like short steps, big feet, little feet, short legs, long legs all with the same result, they fell over very, very well. Even when I managed to get a few good steps with the legs themselves (no body), it was increasingly clear this was not going to happen without some evolution happening in the machine.
One of the major challenges to get a robot walking is finding the right motor. In the 90s motor sourcing was based on luck via salvage or buying gear motor sourced from industrial catalogs that cost more than a typical car payment, which was not an option. I found motors through a surplus catalog and by buying a sample and testing them, I stumbled across a usable torque, a nice slow speed that still left me money to buy a sandwich.
When the motor arrived it had a small wheel attached to it. I planned on removing it, but while I was testing a new idea began to stir in me. Perhaps the thing I was building was a bit upside down. All the motors were on top of the assembly and not at the feet. Maybe I just needed one on each foot that made the feet move forward dragging the legs and body with it.
My first attempt was comical. I created two stick legs and attached them to a small platform that were mounted to feet with a single bolt to allow to and fro motion. I would power 1 motor then other in sequence. What actually happened was similar to learning how to roller or ice skate. Your feet get ahead of your body and you end up on your hindquarters. This was going to take some more thinking.
I reasoned my first failure was due to the lack of stopping the motors quickly enough along with letting them progress too far each time. I had to think in terms of baby steps. I added some micro switches on the pivot part of the legs where they joined the feet, each attached so the motor would only move the foot a couple inches and shut off. This worked well and I started to gain some confidence.
I still had the platform attached on top that was free to swing anywhere, however I needed the platform to stay level with the ground and feet if it was to be stable upright. This turned out to be just a matter of common sense. I added a parallel leg upright on each side to make the platform follow the feet exactly. They were mounted side by side with just a small space between them. From there I quickly got a set of legs walking. It is very amusing to me to watch just a pair of legs walking around by themselves. I managed to automate the steps using some basic circuitry that allowed forward, reverse and turns via radio control.
No sooner had I finished these legs I put them aside and went back to the six motor system... Just being stubborn. I worked on that for ten years off and on... and it still was never anywhere as close to stable as the original design with just two motors.
In 2009 I went back to the two motor design and build some upright walking zombies for Halloween. I got out the old leg system and made a nearly six foot tall one that was entirely automated.
Check out The Walking Zombie video
The Stormtrooper body was a stroke of luck. A retail store had marked them down to $20 so I got one and proceeded to cut it up and apply the body shell parts to the leg mechanism. It is actually a rather grizzly affair. I cut the lower half of the body (waist down) into no less than 12 parts so they could be mounted to the moving legs. The original feet were cut off completely.
Check out The Walking Stormtrooper video
The legs support system can be 3D printed or made from acrylics. A set of roller switches constantly monitor the position of each foot and control the gait sequence for steps. The system is designed to be mechanically and instinctively stable allowing it to carry heavy upper body weight with no risk of tipping over.
I mounted a LCD connected to a micro that resembles a badge on the chest as well as a battery voltmeter to report the battery status. The LCD says Stormtrooper #143256 and the letters scroll across the display. I dearly wanted to get another body and make a C3PO (scratch build from a Stormtrooper body) too but they were all gone, and I am out of money for now anyway.
My collection of walkers has gotten larger and I have built numerous types, numerous body styles and even a few more monsters. They continue to amuse both me and casual onlookers. If you take these out into the public view you get a big crowd fast. The really wonderful feature of these walking mechanisms are that they are pretty cheap to build and those gear motors are still widely available! I continue to revise and make the system better.
Mark Miller's passion is to design and build robots, androids and automation. You can see what he is working on by viewing his You Tube Channel.