Project ShowcaseHave a project that you're dying to pimp? It doesn't matter if it involves robotics, RFID, HCI, or what, as long as you think it's cool, pimp it here! Need more space for pics/videos email us:)
More info on the Leviskate may be found at www.electricunicycle.com as footage is generated and additional documentation is generated.
Hi Guys,
Well, the cummulation of a couple of years of late nights has paid off. Most of the work was in developing a 60V 2KW tilt/controller and motor driver. What they say about designing your own motor controller is true .... DONT! (unless you have 2 years, money, and enjoy radical explosions and fire)
Basically, it's a teeter-totter servo on a go-cart wheel. The wheel provides lateral stability and turning via tire deformation. An inertial system provides forward backward stabilty through wheel rotation. A PID servo monitors the error from the balance point (vertical vector from earth) and rotates the wheel in the propper direction to maintain balance. It makes correcting decisions 200 times per second. A Kalman filter merges an accelerometer and a gyro as niether sensor alone can provide reliable tilt information at the rates needed. The mechanical system employs a balanced arangement of motor, wheel and tappered batteries. Chain drive allows correct gear reduction for necessary torque. Torque counteracts tilt. Torque is acceleration, the deriviative of velocity. Velocity is proportional to motor voltage in the armature.
Trossen FSR sensors provide safety activation and deactivation. With BOTH feet on the platform, rotating through vertical engages the vehicle. When BOTH feet are gone, the rider is missing and active braking (motor shorting) stops the riderless mass of momentum quickly. Obviously, no air tricks with this heavy machine.
What a cool exercise in minimalism. Now you have gone and pushed it to the limit. The only place to go from here is zero wheels! The hover board from Back to the Future
Thanks for the submission and kudos to a great effort on a cool idea. I would have gotten as far as bolting a skateboard to a rolling pin and then given up when I stood on the driveway not moving...
Thanks Jon for adding the clip.
Thanks Alex and Matt for kind words.
There is much question over patent and liability issues bringing such a product to market. I am trying to gather interest. If that breeds financial backers that can push through these issues, production may become a reallity.
For now, the mechanical contruction is rather elaborate - every piece was hand cut/welded/machined by me. Until I can make the mechanical structure quick and easy to build (hub motor comes to mind) I won't be banging out custom complex protos due to time constraints. I did solve the whole electrical side of the project with my 4-layer double-sided SMT board I layed out. It can auto-adapt to brushless operation with the setting of a bit in the uC that reconfigures the commutation logic in the PLD. It packs the tilt controller too and fits in the size of a DVM. By the way, earlier versions were done by etching my own circuit boards. 1-layer smt boards are easy, and can be made in only a few minutes with a laser printer, laminator, a sponge and some etchant.
Right now, I am creating a re-bar rib-cage to replace the wood/carpet fender that I now find ugly. I am increasing the clearance of the board 1 inch. This will allow steeper grades to be climbed as well as get rid of scraping when turning sharp. Did you hear it in the video? As a free bonus, it will also allow a stronger moment arm on the tread permiting sharper turns. Coolness factor goes up because the board appears to hover more than ever. At the same time the wheel that extends above the top is reduced - making for a sleeker fender.
I will post a day-time video when all the enhancements are made (Sat or Sun, not sure)
Hey Alex,
Thanks for asking. I have modified the fender with the re-bar and rubber mat and elevated the board and added more clearance between the motor and the wheel. That works great and the turning is better than ever. The control loop wasn't tight enough to handle bumps at any real speed. Imagine the instantaneous torque that is required to keep a 250lb 6'6 giant upright when he hits a pothole at 10 mph! It is a two-fold problem. 1 speed up the control loop. 2 have more reserve torque. I have located a gear motor that can trade some not needed (above 10mph) speed for much needed safety torque. I will have to weld that in. Fortunately, I found some phase lag in my sensor filters - by reducing that, I can tighten up the control loop. That makes the board feel tighter (on the bench and hopefully when I ride). I have added many more safety and diagnostic features as well. Here is an important lesson learned -- don't skimp on the bulk bypass of a motor controller. The voltage dip that came from an instantaneous torque request reset my micro (forcing an un-planned dismount). So, in summary, I have been busy identifying problems and resolving them. Indeed, it is one thing to make a self-balancing vehicle, and something entirely different to make one that is SAFE and works well. If it stops raining here, maybe I can get some day footage. I will probably postpone rough terrain footage until I install the higher torque motor and transmission. Best regards, Rodger
Leviskate - Self Balancing Motorboard
Linear Mode
