My last email did not go through dev-sis list as the attachment exceeded
the limit. So I have uploaded the attachment on dropbox and sending it again
_______________________________________________________________
Hi All,
I have completed the implementation of Random Agent Movement and Hybrid
Agent Movement. I have processed open street map data to convert it into a
routable network and have used this network for the above two types of
movements. They have been implemented as per the algorithm below:
*Random Movement*:
Input: Source, Destination, inp_time
1. curr_locn = Source, curr_time = 0
2. guess a path from curr_locn to Destination
3. if time(path)<inp_time - cur_time - constant(a parameter - say 5)
choose a random direction
got till next intersection
update time
goto step 2
*Hybrid Movement:*
Input: Source, Destination, inp_time
1. curr_locn = Source, curr_time = 0;
2. Guess a path from curr_location to Destination
3. Choose any two points at random along the path - p1 and p2, where p2 is
closer to Destination
4. if (inp_time - curr_time > time(p2, destination) + constant (say 10))
Learn a different longer path between p1 and p2 and go to p2
along that path
update time, curr_location
goto step 3.
I have tested the above two movements on my local system.
I am now back to my University and there we connect to internet via a proxy
server, which has the ssh port to external machines (port 22) blocked by
default. So I have not been able to deploy the code on the server. I have
requested to unblock the proxy and am expecting it to be done in a few days.
For now, I have run a few tests on my system and attached the screenshots
with this mail. You can find them here:
https://www.dropbox.com/s/mhfd5pup6zdvln4/screenshots.zip
Please note that the input time in pic random-2.bmp is less than the input
time in the pic random-3.bmp, which explains why there is more random
movement in random-3.bmp compared to random-2.bmp, although both have the
same source and destination.
Please let me know of the next steps of the project.
Thanks,
Nadeem
