Tilt - A Balancing Robot (#1) - Thinking about it

Tilt will be my next robot with a focusing on implementing Dynamic Control in Robotics.

There are a good number of videos on YouTube from 2 years ago on how to build a "self-balancing robot". I do find these fascinating, and I would like the experience of building one - of course as cheaply as possible with as many reused parts as possible.  ;)

I am challenged by the engineering aspects of this type of robot.  Perhaps because ... that an inverted pendulum was offered as a senior project when I was an undergraduate. I passed on it due to the complexity at the time, and lack of good motors and controllers. Yes, I am a bit haunted that I *should* have done this project. So, I feel like I should re-live my Control Engineering days and build this robot! :)

I feel that these type of control will be standard in most future interesting robots. I sense that we are still at the early aspects even though Dynamic movement has been a part of the current stable of robots. we are starting to see combinations of quadrupeds -and- wheels. (Soon, I would not be surprised to find reaction wheels also included for more stability and agility.)  

However - after all that is mentioned, these types of developments are still very uncommon in the DYI Robotics space!

Moving forward with this robot, the numbers and types of parts are limited.  Here is what I am considering at this point:

• 2 wheels, recycle, reuse, or print
• 2 motors - 9-12v from a recycled printer
• 2 belts - to drive the wheels and do some gear reduction/torque
• 2 encoders - used to measure the output of the motores
• 1 motor driver for DC brushed motors (L298N)
• 1 micro controller - EPS32 or Arduino Nano or other
• 1 motion sensor - MPU-6050
• 1 battery solution - drives the motors, powers the electronics

What else would I need?

OK - Time to scavenge for parts!  and think about an encoder solution.

Intergalactic Planetary (gears)!

Just doing some thinking and sketching on how best to use Recycled Printer Motors in my robots. 

One of the Totally Not Evil Robot Army motivations, is to make robotic creations by recycling parts from existing technology. To RE-USE, is the best approach to waste management. -- and also of course, because I am really, really cheap.  ;)

Recycled Printer Motors

Most computer printers have these nice (DC Brushed) Motors, but they tend to be too fast and with out a lot of usable torque for my projects. This would be expected, since the printers they come from have a lot of built in gear reduction.  For power, they typically require 12-18v. These are not light either, they have some weight to them. In order to utilize these printer motors, they will need a gearbox to both reduce speed and increase the torque - which is exactly what gearboxes do for motors.

The Planetary Gear is a good choice for a compact way of converting the motors output into a stronger more useful force.

The Planetary Gear is made up of 4 pieces, The Sun Gear, one or more Planet Gears, The Carrier (connecting Planet gears) and on the outside, the Ring.

Named Parts of a Planetary Gear Box

The configuration of a planetary gear box can vary based on specific needs. Any one part of the system can be 'fixed' which allows for the other two parts to become the Driver (input) and the Output.  

Example:  by fixing the Ring and attaching the drive motor to the Sun, it makes the Planetary Carrier to become the Output and move.  Likewise, you could fix the Carrier and attaching the drive motor to the Sun Gear, would cause the Ring to be the output and move.

Here is a quick 3D print or sketch of the Planetary "Wheel".  Here the motor and Carrier are the same part. This allows the motor to directly drive (input) the Sun gear.  Since the planets are fixed, the Ring will move. I have built the Ring gear to be the inside of a wheel.

Motor Mounted on the Planetary Carrier

On the reverse side you can see the planetary gears encased in the wheel. The Sun Gear is removed.

Complete "Planetary Wheel"

Gear Reduction with Planetary Gears

The actual gear reduction and torque increase can be calculated based on the number of teeth in the gears or the radius of the gears (given a specific modulus).  The equations are similar to basic gear equations, however there are differences due to the planetary gear set.  here they are:

For the gear ratio of a basic gear 

    i = Z2 / Z1 

where Z are the number of teeth on Gear 1 and Gear 2

For the Planetary Gear, we have the number of teeth - Zs for Sun Gear and Zr for the Ring. The number of teeth on the carrier does not matter, since it is derived and removed from the equations. (also, it does not matter how many actual planet gears you have).

the equation for planetary gear reduction is:

    i = 1 + (Zr / Zs)

Here is a nice little summary video of deriving the gear-ratios of Planetary Gears.

(Update:  here is a nice summary sheet from the MIT FAB LAB)

Testing the 3D printed gear box

I was able to make this wheel work pretty much out of the print.  I did have some difficulty in the motor shaft 'stripping out' the inside of the sun gear if the gears were blocked.  this is the same problem encountered on the Milli project.  I believe some glue may help hold the gear in place, but this does not seem to be a sustainable solution. I would have difficulty replacing the gear when needed.

As a result, The Sun gear would slowly lift off of the motor spindle when operated at full speed.  (A failure you would not want if the Wheel was part of an inter-planetary rover.)

Looking forward, at these brushless motors, I believe a belt and tensioner system may work best. Most of the motors I have, already have a nylon belt gear on the motor spindle.  I will try to use this combination in one of my next projects.

Postscript

Hi all, this post was somewhat delayed due to some personal health issues.  I am doing much better now, thanks!  I have some upcoming work on a balancing robot - using Printer Motors that I am currently considering.  More to Come!  Thanks! Doug

Two Year Anniversary

It has been an interesting year, who would have thought!

It is now Two Years since I started my Totally-Not-Evil-Robot-Army!  

For this blog, I have had over 14,500+ visits.  Hopefully, this is a sign of more Makers becoming interested in the building robots.  

However, I am savvy enough to understand that many (possibly evil - Internet software Robots) have driven some of my numbers.  Web-Crawlers and Search Engines regularly pass through the Blogger collecting information, pictures etc. I have been directly crawled from an Israeli base search engine responsible for at least 3k hits in the last year, by my estimates. Also, this humble blog had been the target of some Ukrainian SEO- Search Engine Optimization hackers, trying to make comments that cross link URLs and drive up results for their customers. I have removed their posts, but I believe I am still in *their* logs, which results in a lot of other victims in their bot army cross-linking to my site.  Thus, I have a lot of referrals from random non-robot, non-educational sites.  <sigh> unfortunately this is part of our modern Internet.  And, it also make the Ukraine the most popular country amongst our visitors.

The Covid-Crisis has also brought in a number of visitors. Perhaps people are building robots while in lockdown?  some of the most popular locations have been from Hong Kong - perhaps a future robot hot-spot?

The year in passing

Well, I did take a bit of a sabbatical for a few months. I had a lot of personal work in relocating to a new home, and even more effort in  starting a new software business.  I am still thoroughly enjoying robot building and will continue to post my adventures in robotics in a public journal. There are many many robots to be built. As we all know, the robots can't build themselves... until they can. ;)

This years robots include a new design for a Robotic Dog - Mojo3.  In addition (as a distraction?) I worked on a Millipede based robot call Milli (original, eh?).  Both of these robots are still in active design stages. Mojo3 is going through walking tests. For the Millipede, I am exploring better ways to drive the robot using recycled printer motors.

Let keep building!

Mojo3 - A Robotic Dog (#10) - New Video Out

 New Video Out

There is a new Video on YouTube showing Mojo3's first (stumbling) steps.  Check it out to see what it is like to try to tune this little robot in both physically and with software.

Only 9g Servos

The biggest challenge with this little robot is trying to only use inexpensive 9g servos.  These are great hobby servos, easy to use and code, and inexpensive if you burn one out. They do have their challenges in the amount of force that they have. I found this to be restrictive in my first attepts with the original Mojo robot dog design.

I believe with good design and engineering, it should be possible to build a little quadruped robot with them. You can see in the video that my in my first tests, the rear servos were struggling to meet the position commands issued by the controller.  The front and back "hips" and "knees" were given the exact same position commands literally using the same computer array of positions.

Then I changed the rear "hip" leg gear positions. I rotated the rear/hind "hip" leg out by about 30-40 degrees. This was easy to do in my new leg design. I simply pulled the hip off of the mount, rotated it back and put it back on.  There are marks for calibration visible in the pictures. The result was a much smoother "hip flex" motion.

New Hind Leg Configuration

The next step in developing the walking gait is to explore how each leg should move to proppell the robot forward. I am starting with just basic movements of the hind/rear legs.  I have replaced the front feet with "training wheels" (like a kids first bicycle). Now the hind legs can attempt to walk without any resistance to the front legs.

More work to come! 

 

Mojo3 with 'training wheels' for front feet.

 

PS:  Quadrupeds on Wheels is going to be a thing!!  Check out this video of Hybrid Locomotion by ETH - ANYmal.

Mojo3 - A Robotic Dog (#9) - Ready to Program the Step Sequence

Mojo3 is a 3D printed Compliant Quadruped Robot Dog

The 3D printing is done for this build. I am currently working on updating the Arduino software to control the robot leg sequence.  I need to add all the electronics and batteries as well.  More to come!

Mojo3 - Robotic Dog - ready for electronics

Mojo3 compared to Mojo2