[chbot] 6 legged walking Crab

[Richard Jones]

Maybe Sashin has proposed what could be our first group project?
I would be quite interested in having some kind of n-a-pod of my own to
play (oops learn) with.
I can knock out prototype PCBs at cost, although http://www.arduino.cc/
would do nicely.
I wonder if we have someone who could cnc the legs and bodies for a modest
price?
Morris has already recommended a low cost servo supplier.
What else do we need?

Richard

On Mon, 3 Aug 2009 06:25:39 +1200, Charles Manning
<manningc2 at actrix.gen.nz>
wrote:
> On Saturday 01 August 2009 14:56:24 sachin surendran wrote:
>> Hi,
>>      I am building a 6 legged walking Crab with 3 servos on each leg.
>> I am looking for hobbyist who would like join along in the process of
>> building it.
>>
>> My rough plans are :
>> - To perfect the walking algorithm for 6 legged crab with 3 degrees of
>> freedom
>> - Experiment on self learning ( The bot learns itself to perfect its
>> walking motion to take it from one place to another )
>> - Mount a camera for purposes of image recognition.
>>
>>  Its been more fun working as a team than alone on this ,
>> So any robotics enthusiast from christchurch who like to join along, go
>> ahead
>> and drop me a mail.
> 
> Hi Sashin
> 
> I think it would be a good idea to spend some time at the next meeting
> doing 
> some brain storming on where this could go.
> 
> May I be so forward as to suggest that this project can easily be broken
> into 
> distinct projects that can be pitched at different interest areas:
> * Building/programming the platform (ie. the mechanics, sensors, etc).
> * Analytically figuring out walking algorithms.
> * Self learning algorithms.
> * Video processing etc.
> 
> I am particularly interested in the self learning bit. How deep do you
want
> to 
> go with this? ie. Do you want to start with raw servo commands and get
the 
> robot to learn how to wiggle its legs and coordinate them to actually
learn
> 
> to walk or do you want to do higher level functions such as maze
following?
> 
> Or both?
> 
> What ideas have you considered so far? I guess at the most extreme you
> could 
> consider a genetic algorithm approach to learn to walk which would be
> pretty 
> interesting. There are some examples of things like this on the www, such
> as: 
> www:http://eprints.ecs.soton.ac.uk/2955/
> 
> There have also been some interesting mechanical/algorithm co-designs
where
> 
> genetic algorithms have designed both the legs and the gait.
> 
> 
> -- CHarles
> 
> _______________________________________________
> Chchrobotics mailing list Chchrobotics at lists.linuxnut.co.nz
> http://lists.ourshack.com/mailman/listinfo/chchrobotics
> web site: http://kiwibots.org
> When replying, please edit your Subject line to reflect new content.