Last update June 30, 2003

# ICFP / Message Board

Difference (last change) (no other diffs, normal page display)

Changed: 20,21c20
 Nic Tiger.
 -- NicTiger?.

 It's a zip file on the cvs, under the OPTCURVE folder. I'm trying to make the driver/tracer but so far it's been difficult to make it run through one course Maybe we can finish this today. No sign from Burton yet. Nic, I understand your first idea but I think I'll not have the time to do this. About the second I thought your paths were already reduced. In your tracing, how can I decided which points form a straight line, because using double coordinates we can get (100, 110), (300, 110.002), (300, 109.9987) ? Are they forming a line? How large should be my sigma (or delta)? I'm already using points as navigation points, but there are issues we have to deal with, like how fast we can get to deal with curves and so on. -- DanielYokomiso?

Might also be useful. -- HelmutLeitner

### Jun 30, 2003

Hi Daniel, Helmut, Burton!

How is it going? I'am at work now, but maybe I will have couple of hours at the end of day.

Have you succeded to drive car on given path? What are our times? Does sharp corners really mean greater times even with smalle path length?

I have ide about finish-path optimization. We need process first optimization and then iterate the last point of PathCurve? through all pixels of finish line. We should each time optimize path again and thus we will have the optimal curve with optimal finishing point. Do you understand me?

We should also perform point reduction on PathCurve?. When we will have small number of points (I think 200-300 instead of 5000), we can apply basic Auto-pilot system, where each point is NAVIGATION point we should reach. When we reach it, we take the next navigation point and so on. At each moment we have out position and navigation point, so we have distance and direction. The rest should be driving car.

Excuse me for not being able to help in details,

I've sent this night ZIP file to Daniel. It contains sources and binaries to produce optimal paths. Did you get it? -- NicTiger?.

Nic, sorry I got nothing. Daniel sent an e-mail where he talked about an attachment and an additional hint directory in the CVS. I cound find neither. I've also no contact to Burton. I suppose he will turn up in the end and present his super solution. I have no information about traces and simulations at all. I never saw a trace-file from Burton, nor from you. I would like to help but currently I've neither working code nor certainty in information. Where you able to use my manual path suggestions? Have you produced straigtened paths from them? Can you let a car run along a path? -- HelmutLeitner

It's a zip file on the cvs, under the OPTCURVE folder. I'm trying to make the driver/tracer but so far it's been difficult to make it run through one course Maybe we can finish this today. No sign from Burton yet. Nic, I understand your first idea but I think I'll not have the time to do this. About the second I thought your paths were already reduced. In your tracing, how can I decided which points form a straight line, because using double coordinates we can get (100, 110), (300, 110.002), (300, 109.9987) ? Are they forming a line? How large should be my sigma (or delta)? I'm already using points as navigation points, but there are issues we have to deal with, like how fast we can get to deal with curves and so on.

-- DanielYokomiso?

| | | | | | | | | |

(date of last change: June 30, 2003 16:00 )