Emanual Version2, Ready for the Playa

Board Picture from Maker's Blog

Burning Man is a unique event that exposes both people and projects to pretty extreme conditions. 2006 was my second year to attend, and I was fortunate enough to be amongst a creative groups of people who brought a giant mechanical spider to fruition. But I had a vision of a different kind of vehicle for that year. What I wanted to make was a self balancing extension ladder. You could ride it around the desert while standing on a rung, and then when the dust storms kick up, you hit a button and whuup!, the ladder extends and lifts you right above the low-flying dust clouds. Sounds fun? I thought so. But 1 week before scheduled departure, I didn't really have anything prepared and couldn't for the life of me find a cheap extension ladder. So I removed the balancing circuitry from the unicicycle project, modified the code to include differential steering functionality, attached some modified 16" DC hub motors to a piece of plywood and this is what I got:

Board_Playa_Small (12K)

Something quite remarkable happened from when the original emanual got upgraded to a hub motor version with a real rate gyro, it became rideable! Once the gain feedback was properly adjusted, anyone could stand on this platform and be stable, no learning curve at all. I didn't have time to do a proper lean actuated steering like the first emanual, so there is a small potentiometer on the end of the handle that the rider uses to steer. Anyways this worked quite well as a means to get around Black Rock City, and the most remarkable part of all was how little energy it used. With a DrainBrain fitted on board my average current draw at jogging speeds was about 2.5 amps, meaning I could expect 3 hours of use off the single 36V 8Ah battery pack.

The board was taken apart shortly after returning from the trip, and unfortunately I didn't get a single picture or video of my own either from the construction or during its use. If there is anyone out there who did capture footage of this board in action and wants to share it with me, please send me a note ( justinle AT interchange.ubc.ca ).

Worlds First Power-Assisted Unicycle?

One transportation mode that I consider way overlooked is the unicycle. It is by far the simplest and most elegant wheeled vehicle, fun to ride and portable enough to carry on the bus. A few pioneers have establed mountain unicycling as a respectible sport, but the idea of using a unicycle as transportation, to commute to work, well it's not so common. Those who do try to get around on a Uni inevitably ride with a massive 36 inch wheel to achieve appreciable road speeds, but then it's no longer portable anymore.

Power Assist

My vision for the ideal unicycle is this: Gearing between the pedals and the wheel so that only a small 20" tire is needed, and electric assist. The electric assist would have a balancing algorithm so the unicycle is easier to learn, and it would regnerate so that when you go downhills you aren't absorbing energy with the knees but letting it flow into a battery pack. Plus, it would help you go fast, both on the flats and up the hills.

Anyways, while there have been a few people who have made self-balancing unicycles that you just sit on and ride, as far as I know there were none out there that you could pedal as well. So when the VEVA REV! show was approaching again June 2006 I knew it was time to kick this project into high gear.

EUni Frame Ideas (5K)

At first I was considering whether to simply modify a child's bicycle frame into a unicycle, making use of the existing drive chain and attaching a seat to the steering tube. But luckily I was in touch with Paul Peyto, a former UBC student who ran a custom frame building business on the side. He said it wouldn't be too difficult to build up a custom frame and a Sunday afternoon at his shop proved tha was correct.

Control and Power Circuits

Control_Circuit_Uni (99K)

Power_Circuit_Uni (104K)

Frame

EUni_Built_TN (63K)

EUni_End (107K)

Videos

TN_Euni1 (4K) Electric Unicycle at the Spanish Banks VEVA REV! 2006 Show
TN_Euni2 (2K) Electric Unicycle at the Spanish Banks VEVA REV! 2006 Show, turning is the hardest part, for sure.

The Original Project Emanual

The Almost Self Balancing 2 Wheeled Electric Skateboard http://video.google.com/videoplay?docid=8740966685730356244&hl=en

Background

The inspiration for this project naturally came from Dean Kamen's Segway. Dean had the farsighted vision of small lightweight Personal Electric Vehicles (PEV's) redefining urban transportation. Unfortunately, he tried to plan out and orchestrate such a revolution top down with a novel self-balancing scooter. Early media speculation and tight-lipped secrecy made for a ridiculously over-hyped situation, so when the Segway finally hit the stores it was as much a subject of ridicule as of awe. I think though that time will prove Kamen right. PEV's will dominate the roads in due course, only this transportation revolution will not be planned out by either governments or companies; it will be driven instead by grass-roots interest, from people finally awakening to the absurdity, cost, and inefficiency of cars in an urban environment and hopping onto electric bikes and scooters.

Anyways, the idea of reducing the number of wheels on a vehicle and using electronics to keep it stabilized is an attractive concept. Originally I was planning to apply this to a conventional unicycle in late 2003, but put it off till the summer when there would be more time. Then after getting the parts together around August 2004, I saw that Trevor Blackwell had just beat me to the punch with his Eunicycle. But by then I realized that a balancing 2-wheeled skateboard would offer a similar challenge and be far more cool.

Early Progress

The design got started in September 2004 when Don Clarke and I thought we could do this for a self-directed senior project course at UBC. Don sourced most of the necessary components and produced a nice CAD design of the mechanical layout over the ensuing months.

Computer Model of Board Design

In January 2005 we decided to build this device for the Engineering Physics Ball Model, after getting turned down for the project course. Only catch was that the ball model is displayed during the first week of February, so that meant working in high gear!

After about 2 weeks of shop work, we had most of the mechanical structure finished with a remarkable resemblance to the CAD drawings. After this, Matt Chudleigh and I spent a solid 6 days and 2 nights implementing all of the electronics and control systems and working out the kinks.

Mechanics_Complete (55K)

Power Source

There are two 300 watt DC motors mounted on top, each of which independently drives one wheel via a 25:11 chain reduction. The motors were relatively inexpensive for their ratings and were purchased from Oatley Electronics

The original power source was a nice pair of ultra-lightweight 14.4V Lithium Polymer Batteries, each with an 11Ah capacity and a peak current output of 100 amps. These were wired in series for a ~30V source with a theoretical range of about 20km. unfortunately, the battery management circuitry on both packs fried after a few days of testing. That left us using lead-acid batteries in a backpack. Fortunately though the manufacturer has mailed us replacement circuitry and we hope to soon be up and running with lithium polymer again.

A sad looking Lithium Battery Pack

Tilt Sensors

The Board angle is measured with a pair of proximity sensors that detects the distance from the front of the deck to ground. We used two sensors for redundancy. One works by sending out ultrasonic sound pulses and measuring the time for the echo to bounce back, the second emits a beam of infrared light and uses optics to sense how far away the reflecting surface is.

Each sensor type has its pros and cons. The Ultrasonic is nice because it has a digital interface, works off of practically any surface, and has an output that is linear with the distance. Unfortunately it sometimes misses the echo pulse, and it has quite poor resolution at detecting small changes in distance. The IR sensor produces a steady analog output, but the voltage is inversely proportional to distance. It is also affected by sunlight, the optical properties of the surface, and the cleanliness of the lens.

Echo-IR (58K)

The steering control of the board is just like on a regular skate. You tilt the deck side to side. Since the deck is mounted by only 2 rubber bushings, it is able to pivot along this axis by a few degrees. An IR sensor underneath measures this tilt which causes one of the motors to drive faster than the other.

Lean_IR (80K)

Control Electronics

In order to actually drive the motors, we needed a pair of high current bi-directional DC motor controllers. Originally, we had purchased two 20 Amp H-bridge modules from an OEM supplier, but they burnt out in short order about 4 days before the demonstration. That left us scrambling to get an overnight delivery of mosfets from Digikey and build our own H-bridges from scratch. So far these have proven robust and reliable, and don't even get warm to the touch.

H_Bridge (27K)

The actual control circuitry for the board is quite simple. A single PIC16F73 microcontroller reads all the sensor input and commands the H-bridge motor driver accordingly. The algorithm used to keep the board balanced is very straightforward, although implementing it in bug-free assembly code required many long evenings.

Effectively, the microcontroller reads the sensors detect if the deck tilted too far forwards. If so, it increases the motor power to drive the board in from under the rider. The motion is smoothed out by calculating the rate at which the angle is changing, and backing off if it is changing too fast.

Control_Circuit (56K)

In order to help design the controller, we built a computer simulation that models the actual dynamics of our skateboard and predicts how it behaves under different condition. It turned out that these predictions weren't all the useful, because it was not obvious how to simulate a human rider standing on a balancing deck.

Simulink_Model_S (27K)

However, after a bit of trial and error we got it working just in time for e-week. The trick to ride it is to keep your legs stiff so that the controller can sense how your center of gravity moves when the board tilts back and forth.

Results and Videos

Well, obviously the board works! It takes about 20 minutes for someone to learn to ride it though. There are many improvements and modification on the way to solve the many shortcomings in this first prototype, but for now I'll let the videos speak for themselves:

TN_First_Demo (5K) This is footage from about 15 minutes after we first got the board working. It was brought up to the engineering physics club room to show off to all who were lucky enough to be around.
TN_First_UTurn (4K) Here demonstrating the one outstanding virtue of a 2-wheeled skateboard, the ability to do U-turns on a dime!
TN_Chud_Beer (5K) Here is matt zipping down the hallways. Yes that is a beer he is holding.
TN_REV_05 (2K) By the time the Vancouver Electric Vehicle REV! Show came up a few months later, you can see the skill levels improve somewhat! The shield on the front of the board was taped in place to prevent sunlight from confusing the IR distance sensors, which proved a bit of a problem at outdoors demonstrations. Notice also that there are no wires in this video, the battery is under the board and a foot button serves as the kill switch. Hands Free as it Ought to Be.
TN_UBC_BiketoWork (3K) The Bike Co-op at UBC hosts a 'bike to work day' event as one of the many promotional things they do to encourage cycling and bicycle culture on campus. But they wanted to really get some attention this year, so emanual was invited to the show to accompany the various chopper and tall bikes they had on hand.

 

Questions

  • Emanual?
    For the 98% who aren't skateboarders, a Manual is a skateboarding trick when you ride balancing on just the front or back pair of wheels.
  • How Fast does it Go?
    We honestly haven't had the guts to go near full speed, but based on the motor characteristics we should be able to comfortably stay balanced at 25-30 km/hr. Right now I start to loose my nerves about 12 kph, and I think we'll need considerable fine-tuning of the control electronics before it will be safe to ride at maximum speeds.
  • Why do this instead of a regular electric skateboard?
    There is one fully redeeming reason, the ability to turn on the spot.
  • Are the gyroscopes?
    No, just low-cost ranging sensors. Rate gyros would probably help the control considerably since it would let us sense the rate of angle change with way less noise than our current version. This in turn would enable us to tighten up the control gains to achieve a stiffer system.

Here's the Team

Justin_Balancing Chud_Frowning Don_Pondering
Justin Lemire-Elmore Matt Chudleigh Don Clarke

And not to forget our greatest inpisration and champion rider, Madeline Clarke.

Madeline

That's it for now, CYA!

Don_Falling