Date: Tue, 22 Aug 2000 19:03:45 +0100 To: RARS mailing listFrom: Tim Foden Subject: Re: Hello Again At 18:10 21/08/00 -0700, you wrote: >Hello again, > >Just stopping in to say hi. I've lost contact to this list when my ISP >went out of business and then to top it off i redid my whole computer >system so all my addresses and files are sort of misplaced. > >Anything new happening here in the past couple of months? I assume you know that there has been a racing season in progress? It is currently on hold while Gian-Carlo is on vacation. The last race that was run was dated: 2000-07-16. Sparky3 came 1st in both races :-)) (A bit lucky I guess) Current top 6 robot standings: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ K1999 168 pts (France) Remi Coulom Sparky3 81 pts (UK) Tim Foden Felix15 77 pts (Netherlands) Doug Eleveld DougE1 53 pts (Netherlands) Doug Eleveld JOCOSA82 40 pts (Mexico) Jorge Cervantes Bulle2 40 pts (Belgium) Marc Gueury ====================== I think it would be interesting for authors to give a quick explanation of how their robots work, and a bit of the history. Especially the ones not currently publicly available. Sparky was about the 6th robot I wrote. All the others were lane driven, and quickly got very complicated to code, debug and improve. They attempted to optimise the driving path by recognising s-bends etc. The first 3 cars (Mouse1, Mouse2, Mouse3) used circular arcs to drive around corners. The next (Ellipser) used ellipses, and this was a quite successful car. However it still ran off the track a lot. My 5th car used a spiral (Spiro) for the arc, but it was difficult to perform many of the calculations required as the robot ran around the track. I decided that there must be a better way... So I had a look at the "other" tutorials... the ones by Ralph Scott. :) This lead to the development of Sparky. Sparky looks ahead into the distance to find the longest straight line it can find. This straight line will be a tangent to a curve. In essence, Sparky uses this information to plot a course upto and around the next curve. It uses ellipses to generate the driving lines around curves. If anyone has any questions about Sparky please feel free to ask. [I really should get around to cleaning up the source of Sparky and releasing it to the public domain.] ====================== Personally I would be interested in authors quick explanations of bots like Bulle2, Felix15, JACOSA98 and K1999 work. I guess that Bulle2 is similar to Bulle? But I don't understand how it works! Felix15 seems to use some formulas to push the car away from the edges of the track? JACOSA82 look has a very similar driving path to Sparky, but seems to be let down by using circular arcs for corners? K1999. King of the Hill! Looks to use a global path optimisation scheme. I wonder whether there would be some curve that could be used to piece together to make a similar path? Cheers, Tim. -- Tim Foden