Let's Make Robots!

Chair bot

2013-02-23_16.09.53.jpg1.13 MB

What do you do when your 6 year old is excited to share his idea : "Let's build a robot!" Well obviously you start building a robot with your son! At least that is what I have done. This robot will be a humanoid torso on a swivel (office) chair base.


Video of chair base before torso was added:



Video of head turning:

Now with Pan and Tilt!


(old) http://www.youtube.com/watch?v=WxfH0qB_MCo&feature=player_embedded



Build Blog:


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Six year old are not much different than 40 year olds...other than having no life experience to dampen their enthusiasm.  I was excited that my boy wanted to build a robot, but was nervous that his expecatations of an autonomous C-3PO might be a bit beyond my abiliites.

I started him down a path that would ensure that we were on the same page, and that I wasn't agreeing to something that wasn't going to come to fruition.  So the first thign we did was "design" what the robot would look like and how it could move.

So with my expert art skills I sketched out his description while he told me how he wanted the robot to move.  I offered a little guidance, in order ot avoid being commited to a 6ft bi-ped that could take over the world.  This is what we came up with:

Once we had an idea of the direction we were going, I suggested that we make a mockup.  This would give a really good feel for size and work out some basic issues.


So after a couple of build sessions we had a card-board and PVC (non working) "robot"

After building the prototype, it became clear the base was going to need to be a bit more streamlined.  We had put casters on the four corners, and that inspired me to search for a base of a chair.  As luck would have it, about a day after asking our maintainence crew for a broken chair, I had my donor.


We were able to present our robot at the Mini-MakerFaire in Rhode Island!


During setup one of our drive batteries shorted out and fried.  So we were not able to demo how it moved.  But we did have video to show at least.  We were able to demo the head movements, so that was fun.  It generated a lot of interest and I left with a very tired jaw (from talking to everyone wanted to engage)


On the way to the MakerFaire

Ha! I like it!

After frying a battery last year just before showing the robot off, I decided a more robust and safer power system had to be setup.  This system now has a single power source, main power switch, emergency power disconnect, and switches for each sub-system.  It also gave us an opportunity to add some much-needed bling!  You can see more of the details at my Half Built Blog.


Power On and Regulated

I have created a simple web interface that allows me to control the servos through the Beagleboard.  It's been fun to see all this come together.  The power system has been key.  I can switch between a transformer or batteries at will and switches and digital voltage display have been a great way to monitor what's going on with power and I have been able to "see" how one system effects the others (It's clear I need better isolation of the modules).


Anyway I have replaced my simple head turning with a pan and tilt mechanism.  It's pretty brute-force right now.  The "turning" servo has a simple platform connected to its control horns that holds the "tilting" servo.  The tilting servo has a bracket that tilts around it. 

This bracket holds the video camera and eventually the helmet.  With just the bracket it's pretty smooth.  But with the camera on it's pretty choppy.  You can really notice the torque after a quick turn.  Tilting up is slower than tilting down.  This can be balanced out a bit, but will cause the turn-torque to be greater.


Additionally I am using cheap SG90 tower pro servos.  So there is a lot of jitter.  I am wondering what the best option is here.  Ditch the direct mount Pan/tilt mechanism, upgrade the servos, or both?  

Here are the latest videos:

UI + head move + commentary :) (2:00)

UI + head move (1:00)

UI + speed control + power + head move (1:45)



Since I can't see the details I'm just guessing. I did do a pan/tilt for a camera many years ago and these worked for me.

I'm thinking perhaps 3 things.

1. Upgrading the servos isn't a bad thing. Underpowered motors can cause jitter.

2. For smoother action, perhaps some ball bearings in the bracket and the servos.

3. Perhaps a PID loop for acceleration and deceleration would make the end-points smoother.

However, it looked like you had a lot of jitter in be center. That would be the dead band in the servo. Perhaps with a digital servo that could be dealt with, but I really don't know. I thought the dead band was supposed to protect from that sort of jitter.

A question: how often per second are you updating the servo position?

Once my 3d printer gets here I get to do these experiments and I get to see if my own advice is any good. :)

This is coming along very nice.  A very interesting project.  I can't count how many old office chairs I have pitched over the years.  Never even occurred to me what a nice robot base those would make.  Good luck with it and please keep us up to date.

So I now have color tracking working on board the beagleboard computer.  I also have servo control (via Torobot servo controller).

However, when I try to run both the USB camera and the USB servo controller together at the same time the results are aweful.

I can usually make about 3-5 servo calls before the system resets.  If I don't connect the camera I can control the servos for hours. 

If I don't use the connect the servo controller, the color-tracking will work till I tell it to stop.


But as soon (or shortly after) as I use both, the computer restarts.

I'm wondering if the servo controller (or servos more specificlaly) is causing a noisy power supply and making the Beagleboard trigger it's protection circuit (causing it to reboot).