DIY Segway-Style Balancing Robot

clarionhaze writes "Many have tried, and failed, at getting a robot to sustain it's own balance. However; Steve Hassenplug accomplished it with with a small robot he made out of legos and a program in C that runs on BrickOS, an OS made for Legos! You can check out his site or read the article over at TechTV." Update: 01/18 15:52 GMT by T : Unanimous Cow writes "David Anderson of the Dallas Personal Robotics Group has an excellent web page with images and movies of his two-wheel balancing robot. This one uses a single-axis inertial measurement sensor and is very robust on uneven surfaces and off-road."

Comments

not news...

[IdahoEv]

"Many have tried, and failed, at getting a robot to sustain it's own balance

Many have tried and succeeded, as well. Balancing a two-wheeled robot (or balancing a pole from the bottom, or a four wheeled robot on top of a randomly rolling cylinder, etc.) is a fairly common design project for undergraduate engineering students in control theory. I'm not surprised someone did it in legos; they're a perfectly good platform for such an experiment.

Kamen was not the first to come up with a balancing machine -- he's just the first I know of to market a useful (?) consumer product using such a system for human control of a vehicle. One of those head-smackers ... "why didn't I think of that?".

for the record

[paRcat]

I've read slashdot everyday for a few years, but for some reason never saw the original article. And given my interest in robotics, I'm very glad this was posted, repeat or not.

I don't understand the mentality of someone who feels the necessity to point out every mistake that slashdot moderators make. I mean, you could be a troll, or you could just be anal. In either case, you contributed nothing to anyone. You apparently think someone has hired you to act as a critic. Critics annoy me to, unless they happened to be named Homer Simpson... then I just laugh.

Self-balancing unicycle

[Animats]

Two-wheeled dynamic balancing isn't that hard. A simple feedback loop coupled to a sensor that measures tilt angle will do it. (A "tilt sensor" or accelerometer alone won't give you valid tilt, though. But see below.) One wheel, though - that's hard.

A friend of mine built a self-balancing unicycle at the Stanford robotics lab in the 1980s. That's a much tougher problem. There's no metastable point that can be maintained with small corrections.

If you want to do this, the correct sensor suite is a rate gyro and a pair of accelerometers. Back in the 1980s, both were expensive; now they're cheap ICs. They're auto parts. To get a good value for "down", you integrate the rate gyro and run it through a high pass filter, then add the accelerometer value,filtered through a low pass filter.

At My Middle School...

[Poeir]

I spent a lot of my childhood playing with Legos, and the time I wasn't spending playing with Legos, I was playing with computers.

Anyway, to the point, during middle school, one of my projects was to build simple robots and control systems, using Legos and an Apple][e. It's been quite a few years, so I barely remember the details, but one of them drove around, another one acted as a motion sensor; the most complex one undertook a series of actions when the motion sensor was triggered, so it was nothing extraordinary; but Legos are (or at least were) used in some schools. This was a few years before Technics were even available, I think; so they may have even been the prototype.