And Now for Something Completely Different - APM:Bot

Ive been wanting to build a Hanging Pendulum robot for the last 30 years

Technology (and inexpensive consumer parts) finally conspired to let me try and realize my dream.

Inexpensive 500 watt 48 volt 26" Chinese brushless bicycle wheels form the basis of my bot.

And the chassis is constructed from T6 aluminum plate and extrusions with stainless steel side forks and hardware.

The 2 intermediate plates are 3/16" black ABS.

The Bottom shelf is for 4 12 volt wheel chair batteries that total 92 lbs in weight and wil be series wired for 48 volts.

The intermediate bay will hold 2 Kelley reversible 48 volt 500 watt bicycle wheel controllers, a DC to DC 5 volt and 12 volt converter, an APM (later Pixhawk), a receiver, Telemetry and some assorted interface electronics for the Bicycle wheels hall devices and throttle.

The top shelf will support my HP Envy Laptop for Kinect and Laser Scanner interpretation.

A brushless gimbal stabilizing (and scannable) platform will be on top with a Kinect, laser scanner and FPV camera.

As it is, the chassis has worked out very well so far, I didn't manage to build myself into any corners I couldn't get out of and its is as robust as it looks.

Fabrication took more work than I had initially envisaged and most of the layout was simple manual markup.

As it is now, the wheels and bare chassis weighs 63.0 pounds and the battery pack will weigh 92 lb.

There should be about an additional 20 pounds of stuff in the first pass (Vision system but no stabilizers or arms), so it will weigh about 175 pounds. 

This is not a light robot and you definitely would not want to get hit by it.

The 1000 watts worth of motors are definitely overkill, but I wanted the extra torque for operating at low speeds and the wheels I have chosen are very efficient when operating in the 20 to 50 percent power range.

Please feel free to comment and as I continue to make this operational I will post more about this either here or on subsequent Blogs.

Best Regards,


Views: 2770

Comment by Gary Mortimer on October 26, 2013 at 11:36pm

Nice, I think

Comment by Randy on October 27, 2013 at 12:37am

Are you going to use Jason's balance bot code?  He ran it at AVC and it seems to work.  We could drag it into the main repository if there was demand for it (and assuming Jason's ok with that).

Comment by Justin Stiltner on October 27, 2013 at 7:42am

That looks really nice, have you checked to see how the motor controllers behave with slow rotation speeds?  The sensor less wheels I looked at a while back ended up having a pretty quick minimum speed.  Also have you tried running one of those wheel motors from an RC brushless controller just for kicks?  Good luck!

Comment by Gary McCray on October 27, 2013 at 9:52am

Hi Randy, I appreciate the thought and Jason's code might be useful, but a hanging pendulum bot has pretty much the opposite responses to a balance bot. In a balance bot, balance is dynamically maintained and movement is accomplished by unbalancing the bot in the direction of motion.

In a pendulum bot, balance is statically maintained and when you move the bot rotates against the direction of motion, so you need to limit acceleration to that which does not over rotate the bot.

As it is now with my bot chassis it is almost neutral with about a 3 pound top offset, but once the batteries are in place it will be very strongly balanced free hanging.

It will bob about when it is in motion, eg the real necessity for a stabilizing platform for the vision system.

And the last thing you want to do with one of these is slam on the brakes.

And Justin, Even though these wheels came with controllers, I bought a second set of controllers that have reversing and regenerative braking and which are designed for slower vehicles like golf carts (they are also more efficient).

And of course these are sensored wheels with I believe 36 poles and 3 hall sensors each giving 216 edges per revolution.

While the controller will be interpreting these for control, I will also be interpreting these edges myself to determine wheel position accordingly which should yield better than 1/2" resolution on my 26" wheels.

The programmable Kelley controllers current (torque) response characteristics are adjustable up to maximum current (500 watts).

Since these are sensored wheels I do not think they would work well with a non sensored ESC and it is hard to find 500 watt 48 volt ones in any case.

In its initial testing mode I am going to be using the APM to provide the outputs to the controller, throttle, reverse and regen and to input from the hall devices and simply provide link to my RC transmitter.

This should allow me to dial in the controllers and low speed response. 



Comment by John Hestness on October 27, 2013 at 11:56am

I think there is demand for balance bot code. (that was a hint)  Starting an offshoot of the rover project would be one way to think of it.

That is an impressive piece of hardware Gary. I will be following your progress. Keep posting.

Comment by Arthur Benemann on October 27, 2013 at 12:15pm

@Gary McCray

If you need some support from DroidPlanner to have an easy to use tablet interface we are happy to help.

Comment by Gary McCray on October 27, 2013 at 12:21pm

Hi John, I agree that Jason's balance bot code would be a really good thing for us.

For 2 wheel robots, balance bots are actually quite practical for a lot of applications, hanging pendulum ones like mine have a narrower application range.

And Jason's balance and control code seems to work very well.

By necessity hanging pendulum bots tend to be short and heavy and their motion is limited by their normal pendulum response.

Balance bots use their dynamic imbalance to their advantage and preset their angles to allow for optimal acceleration and deceleration. They can also be tall and relatively light and scale well.

Realistically, balance bots probably have a bigger future than pendulum bots.

Comment by Gary McCray on October 27, 2013 at 2:40pm

Revised wiring layout for APM:Bot:

Comment by Jesse on October 27, 2013 at 3:15pm

it's lookin' good mate!!  I would suggest that you either use thicker L section or weld that bottom joint.  95lbs of batteries exert an awful lot of force on the frame going over bumps... I suspect that that L section will fail at some stage as it's a little bit light... how are you securing the batteries on the bottom shelf?

Comment by Gary McCray on October 27, 2013 at 4:15pm

Hi Jesse,

That is 1" x 1/8" thick 6063 T6 Angle  holding a 3/16" thick piece of 6061 T6 aluminum diamond plate on the bottom.

It really shouldn't have any trouble with 92 pounds of batteries, In fact should be good for well over 600 pounds static load.

And since I am not planning on high speed off road travel with this I don't think the dynamic load should be too much of a problem either.

If I was building a high speed off road bot I'd have changed a lot of things, It would have suspension for starters.

I'l be using heavy rubber mat underneath the battery and around the sides strapped in with two battery straps. (It will likely have an acrylic front and back panel as well.)

The bottom line is that this is a development platform and whatever isn't up to snuff will get upgraded and replaced.

At this point I am happy that the basic design seems like it should be functional and it is quite robust.

Best Regards,



You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service