BeagleBone project

[Paul Brook]

Me and Nav were discussing the Farnell/BeagleBone hacking project this
evening, and may have come up with a cunning plan.

For those that haven't been paying attention, this is a competition sponsored
by Farnell/Element14 to promote the new BeagleBone Black. We're in
competition with Pumping Stateion One[1]. For those that really haven't been
paying attention the BeagleBone Black is a new improved, cheaper variant of
the old BeagleBone, which is sort-of similar to a raspberry Pi, but better
(faster CPU, more IO).

Our idea is to do a robotic table football.

Take a regular football table, then add camera and actuators controled byt the
BB, and have it play against a human.

My thinking is that this combination of computational complexity (image
analysis to determine ball position from camera), realtime response, and low-
level IO (motor/servo/etc control) is pretty well suited to the beaglebone.
There's also smaller on-the-side bits like an electronic scoreboard (for which
I have the nice big 7-segment displays!).

The end result is nicely interactice, with enough meat behind it to keep the
real geeks interested.

At the same time I don't think any one part is ureasonably hard, and it splits
up into reasonable independent sub-projects. The image regcognition is 2D,
and we can probably get away with a fairly simple color filter. For the
actuators there are various motor/servo/spring/gravity combinations. A bit of
custom linkage will be required, but nothing too taxing and I don't expect it
to need particularly fine tolerances. The scoreboard is a fairly
straightforward small electronics projects.

Anyone have a spare football table they'd like to donate? If not it looks like
we should be able to pick one up off ebay.

Thoughts?

Paul

[1] http://pumpingstationone.org/

[TK]

Hi Paul
An air hockey table game may be more achievable in the 4 week given for this competion. It would require the computer player to move in x and y. It may also make the computer side algorithm a bit simpler.  My concern is about the lag between image recognition, trajectory calculations and then mechanical action for the football game. Maybe things have moved on since I last looked into this years ago and my concerns are unfounded.
Taimur

--
You received this message because you are subscribed to the Google Groups "Leeds Hack Space" group.
To unsubscribe from this group and stop receiving emails from it, send an email to leeds-hack-spa...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Morgan Barke]

Just a thought...
Perhaps with a air hockey table, you could get away with only the x-axis. If it was reactive and accurate enough, you increase the returning velocity of the puck by hitting the edge at speed.And if the puck is sent at a reasonable velocity it should bounce back just fine...

Morgan.

On Jun 1, 2013 1:10 PM, "TK" <taimu...@gmail.com> wrote:

Hi Paul
An air hockey table game may be more achievable in the 4 week given for this competion. It would require the computer player to move in x and y. It may also make the computer side algorithm a bit simpler.  My concern is about the lag between image recognition, trajectory calculations and then mechanical action for the football game. Maybe things have moved on since I last looked into this years ago and my concerns are unfounded.
Taimur

On 1 Jun 2013 01:02, "Paul Brook" <pa...@nowt.org> wrote:
>

> Me and Nav were discussing the Farnel...

[Stanto]

A simple question but do we have dates for this yet ?

[Stanto]

http://www.mate.tue.nl/mate/showabstract.php/11488

A paper on how to track a ball using a camera on foosball.

Neil made the suggestion of actually making the playing field effectively touch sensitive and tracking the ball that way?

On Saturday, 1 June 2013 01:01:55 UTC+1, Paul Brook wrote:

[Daniel Fligg]

My understanding is that a foosball table has already been purchased, so good luck with only an X axis ;-)

--

[Samwise Wilson]

The ball frequently leaves contact with the table surface in foosball.

Personally I would have IR illumination and LDR system. Just make the players from IR transparent materials and only the ball will be detected by making sets of LDRs low. The advantage is this will give X Y position as a analog electrical measure and so is not limited by your fancy pants camera analysis rates.

[Stanto]

Are you not still going to have to handle how the IR responds when reflected through a transparent material with that ?

I don't know what you mean by an LDR system.

[Paul Brook]

> The ball frequently leaves contact with the table surface in foosball.
>
> Personally I would have IR illumination and LDR system. Just make the
> players from IR transparent materials and only the ball will be detected
> by making sets of LDRs low.

I'd be amazed if that worked. My experience with IR proximity sensors is that
they're extremely finicky. It's hard enough getting good results in controlled
curcumstances with a single axis over ranged of a few inches. We're talking
about two axis sensing over several square feet of playing area. I don't see
how a few LDRs is going to give you anything vaguely useful.

Some sort of light beam detector is probably a viable solution for detecting
when the ball crosses the goal line, but that's about it.

I also suspect "make players transparent to IR" is substantially harder than
it sounds.

Paul

[Samwise Wilson]

Its the same technology that powers certain types of HP touchscreens. They have a bezel with embedded IR LEDs and photodiodes.

[Stanto]

When are people free to meet up and work/discuss this (if people think it needs it)?

Sunday this week's best for me, but I may be able to turn up another day.

[Craig van Vliet]

There is an IR proximity sensor (or there was at least :p) in on of my boxes, it you want to try it for anything. What sort of camera are you using, I have a 60fps global shutter camera from the imaging source, no idea if you can get it working with beaglebone, it took this to make it work in Linux -> http://unicap-imaging.org/.

As always I struggle to get there atm, but I do find time to drop stuff of like the 4pi, so if any of this can be of use, I can try.

On 12 June 2013 15:56, Stanto <sta...@gmail.com> wrote:

When are people free to meet up and work/discuss this (if people think it needs it)?

Sunday this week's best for me, but I may be able to turn up another day.

--

[Christopher Stanton]

Sure, we can give it a go. Thanks!

[Andy Irving]

This may be of use, or may be not, I've never used it http://opencv.org/

[Paul Brook]

> This may be of use, or may be not, I've never used it http://opencv.org/

Yes, AFAICT that's the de-facto standard base library for doing any sort of
[open-source] image processing. Doesn't actually give you anything
immediately useful on its own, the clever bit is how you apply it.

Paul