[Kristian] just put the finishing touches on his full size Segway built from scratch.
Back in 2012, he made a small balancing robot using a gyroscopic sensor and a PID controller — you can see the original post here. The cool thing is, he’s basically just scaled up his original project to create this full-size Segway!
It uses two 500W 24V DC motors (MY1929Z2) on an aluminum check plate frame, with the rest of the structure made from steel plumbing and fittings. What we really like is the steering linkage; similar to a real Segway, you pull the handle in the direction you want to turn. He’s accomplished this by putting another length of pipe parallel to the wheels which is connected by an elbow fitting to the handle bar. It’s supported by two pillow block bearings, and in the back is a fixed potentiometer — when you lean the handle bars one way, the pipe rotates, spinning the potentiometer. To make it return to neutral, he’s added springs on either side.
There’s an impressive build log to go along with it, and a great demonstration video after the break.
He’s even written an Android app for tuning the PID values while driving it!
[Thanks Sigurd!]
Great build! He may be able to just epoxy a strain gauge to the bottom of the frame to replace the dead man switch.
I have actually thought about that. I might try it at some point :)
I actually thought about that. I might try it at some point :)
Nice build!
On some exercise treadmills the deadman switch is a simple scheme where the user has a small plastic-encased disc magnet on a short lanyard or loop, worn over the wrist or otherwise attached to the clothing. This magnet is placed on the control panel where there is a reed switch underneath. If the user steps off (or falls off!) the lack of the magnet of course disengages the mechanism and it stops. You could do something very similar with a magnet on a wrist loop I would think.
We thought about that as well, but didn’t have time to get one.
It might be a nice improvements, as some people that is not used to riding on it sometime releases the button and fall off :)
Nice job. I would suggest, to watch out for cliffs.
I’ve heard if you drive over a cliff on a Segway it is bad Karmen.
good job. this deserves more than 2 comments.
The one thing I would change would be replacing the POT with an optical encoder. Pots are only rated for so many cycles before they start to seriously wear. Optical encoders should last forever.
While I agree that a POT is not the ideal solution, I would advise against optical encoder as well… How many times have you turned a stereo speaker and had it not set the volume where you want? Or better yet, had it not do anything at all?
Unfortunately, I can’t think of a better solution at the moment.
Totally agree with the 2 comments statement. This is such a great hack, I want to see some discussion to match!
An optical Encoder / rotary encoder would work just fine. They are used all throughout industry for the accurate positioning of shafts etc
Usually those are mechanical encoders not optical. Optical or magnetic should last just about forever until either the LEDs/sensors die or they get bashed to death.
You could use a differential hall sensor setup… No electrical contacts to wear out, and it has an analog or digital output based on the sensor chip you use.
The output is absolute I might add, which an encoder is not. You have to derive the measurement in code. The absolute nature of the measurement is something that would be extremely desirable in this context as a matter of safety…
You don’t need absolute for this application. incremental is fine. Also halls probably won’t have the resolution you need for this. Since the range of motion is really kind of small (+/-20deg maybe?) you need a higher resolution encoder to make sure you have enough feedback. If there are 1024lines/rev on your encoder, which is pretty standard for motor feedback, +/-20deg is still only 113 lines of resolution. That’s probably plenty for this.
What industrial motors (especially brushless) use is a combination of low resolution halls and high resolution encoders to get rough absolute position and then go incremental for fine position control.
it’s the lack of comments that imply how good of a project it is!
very nice and very detailed explanation!
Just a edit the motor is a MY1020Z2
Wauw I used to live there! Cool :) It’s nice to see someone hacking away in the dorms!
I want!! That, sir, is pure awesome right there. Why didn’t I think of building one? Have a good one : )
Kind regards,
Aidan