Defining Detour (SOS)

hey guys,
i wanted to know whether there is any system to define a minimised detour in a system of points.. i have tried the delunay edges for the direct path system..

many thanks

Replies to This Discussion

Permalink Reply by Daniel Piker on May 4, 2010 at 12:26am

What do you mean by 'minimised detour' ? Like a minimum spanning tree ? or the travelling salesman problem ?

Permalink Reply by Tushar Sharma on May 4, 2010 at 3:42am

Hey,
I have attached the pictures.. I wanted to use that system in urban plan to reduce the length of the network, hence reducing the area consumed.. I have attached some scans to give some idea..

Permalink Reply by Daniel Piker on May 4, 2010 at 4:10am

Interesting stuff. Where are these scans from ?

I did start exploring a simulation of this approach with threads in water in Kangaroo.
See the video here.

In theory the same approach can just be extended to larger numbers of crossing threads. (When I tried it I found it got too slow when there were large numbers of points all attracting each other, but that was in an earlier version and it should be a bit better now)

Permalink Reply by Tushar Sharma on May 4, 2010 at 4:32am

Hey,

Its from a book called Non Planned Settlements ..Hey i saw the video.. And thats looks like what i want.. Cant understand the algorithm though..Can u help me out with it.. Can u send me and example showing how it works..

Thanks a Lot

Tushar
arch.tusharsharma@gmail.com

Permalink Reply by Tushar Sharma on May 4, 2010 at 8:43am

Hey Daniel,
Can u link me to the example.. I wanted to refer to that.. Its urgent..
Thank u

Permalink Reply by Daniel Piker on May 4, 2010 at 10:03am

I don't have that file to hand right now, but it's a relatively simple setup in kangaroo (which you can download from the link in my previous reply).

Basically the threads are divided into segments, each of which acts as a spring.
Then all the nodes attract each other (using the interconnect component) with some power law. Different exponents give different results but probably around -1 to -3 is good. (The real Cheerios effect has a slightly more complex equation!).
Speed becomes an issue for larger models though because you get a combinatorial explosion of interconnections as the number of nodes increases.

Permalink Reply by Tushar Sharma on May 4, 2010 at 2:56pm

hey daniel,

Threads are divided into segments, i didnt really get it.. how can i do that..Suppose i have set of points connected to each other by lines.. how do i proceed further?

Thank u

RSS