Re: [pygame] Pygame, cwiid and ode - movement

[flx <xf3l1x@xxxxxxxxx>]

2008/8/7 Nathan Whitehead <nwhitehe@xxxxxxxxx>:
> On Thu, Aug 7, 2008 at 2:39 AM, flx <xf3l1x@xxxxxxxxx> wrote:
>> I'm trying to make a foosball [1] game wth pygame, using the wiimote
>> to control the bars.
>>
>> I've managed to control the rotation using the accelerometers, but i
>> can't get the sliding part right. I've been asked to use the ir
>> sensors for the sliding movement, but I just can't get it right.
>
> I think you have two separate questions here:
> * how to get IR working on the wiimote and get IR data
> * once IR is working, how to interpret the data to make the paddle
> slide around correctly
>
> For the first part: which Wiimote library are you using?  Does it
> support IR?  Is it buggy with your particular Wiimote/bluetooth
> combination?  What operating system are you using?
I'm using the libcwiid python moudule (so it's under linux) wich works
well with my hardware, i've written some documentation covering the
basics of interfacing using python-cwiid here [1]

>
> For the second part: I recommend making a simple program that just
> displays squares at the reported IR positions, setting the screen to
> 1024x768.  Then try moving the Wiimote around and take notes about
> where the squares go.  Where do they go when you slide left?  What
> about when you point downwards?  Rotate clockwise?  Then use those
> notes to write code to control the paddle directly.  You might need
> some calculations here (argh, math!).

Doing exactly this i realized that my problems were the reflections
and the points going out of sight, right as you say below

>
> Once you've got this down you will have control of the paddle but it
> will be jerky and flickery.  Also when you move too far to the edge of
> the screen the paddle will stop moving (one IR dot goes off screen).
>
> To fix the jerkiness, a first step is averaging several values over
> time.  A better but harder approach is a Kalman filter.
Didn't kow about Kalman filter (i'm rushing to wikipedia now :) ) but
definetly some kind of estimation will help

>
> To fix the flickery problem: if there are 0 dots reported, use the
> most recent positions.  That stops the input from disappearing for a
> frame.  If there are too many dots (3 or more), then you want to pick
> the 2 dots closest to the last reported IR positions.
>
> To fix the edge-of-the-screen problem, a good approach I've used is to
> guess where the missing IR dot should be.  I assume that the dot off
> the edge of the screen moves just like the one that remains on the
> screen.  This won't be strictly true if the wiimote is rotating or
> changing distance to the sensor bar, but it will keep the paddle
> responding to input rather than stopping abruptly at the edge of the
> screen.
I was thinking of working with the accelerometers when not enough ir
data is avaliable, it definetly won't be nearly as accurate, but could
work
>
> FYI: I'm working on a pygame library that supports the wiimote on all
> platforms, it will be done Real Soon Now.
> --
> Nathan Whitehead
>

Thanks, you've given me some very useful insight


[1] - http://flx.proyectoanonimo.com/proyectos/cwiid/

-- 
.__.·º(foolhu!)
(oO)
/ | | \