From: rars-d-request@lysator.liu.se Subject: rars-d Digest V97 #5 X-Loop: rars-d@lysator.liu.se X-Mailing-List: archive/volume97/5 Precedence: list MIME-Version: 1.0 Content-Type: multipart/digest; boundary="----------------------------" To: rars-d@lysator.liu.se Reply-To: rars@lysator.liu.se (Rars mailing-list) ------------------------------ Content-Type: text/plain rars-d Digest Volume 97 : Issue 5 Today's Topics: Bug in pitting code RE: Track: Jerez Re: Track: Jerez Error in Road America? Qualifying Lap Times Re: Qualifying Lap Times Re: Error in Road America? Re: Qualifying Lap Times Re: Error in Road America? Statistics ------------------------------ Date: Thu, 13 Nov 1997 17:13:21 +0200 (EET) From: Jussi Pajala To: Rars mailing-list Subject: Bug in pitting code Message-ID: Content-Type: TEXT/PLAIN; charset=US-ASCII Hello guys! I checked out the new unix version of rars. Pitting implementation is quite simple but it's working well. At least after I found and fixed one bug... Variable pit_lap wasn't initialized at all, so I put pit_lap = -1; into line 788 carz.cpp, just after line 'laps = -1;'. This seemed to fix the problem. Problem was that sometimes cars didn't stop when they requested to pit. --Jussi ------------------------------ Date: Thu, 13 Nov 1997 16:20:05 +0100 From: Torben Thellefsen To: Rars mailing-list Subject: RE: Track: Jerez Message-ID: <01BCF050.0211EB70@PC6> Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 8bit >> JEREZ CIRCUIT, Site of the 1997 European Grand Prix >> Torben Thellefsen (torben@dir.dk), 12th November 1997 >> Map obtained from http://www.tyrrellf1.com/ >> >> Narrow, difficult track. A good driver must be able to treat up to 3 >> turn segments as one long turn. > >Aren't all of your tracks like this torben?? Well, yes. But please blame the guys who make the real circuits! ;-) Would you rather have simplified versions of the tracks, where the turns are merged into one, not quite so accurate segment? I wouldn't... >This one actually exposed a big flaw in my old car. But blender >makes it around quite well. WappuCar and Bulle rule on this track. And Blender tries to keep up. I actually saw a funny behaviour from Bulle on Jerez. Bulle is usually very cautious when it comes to overtaking - I have a "middle lane" driver that Bulle has great difficulties passing (the same goes for ralph2, btw). But about halfway through the 73 laps race, when Bulle was at 5th place or thereabouts, at least one lap down, it suddenly decided to go for it (having fastest qualification time, btw), and started overtaking other cars like a real pro. And sure enough it won the race! By a large margin. I've never seen Bulle (or any other car) change behaviour like that. Usually cars tend to slow down as they accumulate damage instead of going faster and overtaking in a wild manner (compared to the usual Bulle behaviour). I kinda liked it, btw ;-) Regards, Torben ------------------------------ Date: Thu, 13 Nov 1997 11:10:34 -0500 From: Ralph Scott To: Rars mailing-list Subject: Re: Track: Jerez Message-ID: <346B267A.22F2@netusa.net> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Torben Thellefsen wrote: > > >> JEREZ CIRCUIT, Site of the 1997 European Grand Prix > >> Torben Thellefsen (torben@dir.dk), 12th November 1997 > >> Map obtained from http://www.tyrrellf1.com/ > >> > >> Narrow, difficult track. A good driver must be able to treat up to 3 > >> turn segments as one long turn. > > > >Aren't all of your tracks like this torben?? > > Well, yes. But please blame the guys who make the real circuits! ;-) > > Would you rather have simplified versions of the tracks, where the turns are merged into one, not quite so accurate segment? I wouldn't... Well... > > >This one actually exposed a big flaw in my old car. But blender > >makes it around quite well. > > WappuCar and Bulle rule on this track. And Blender tries to keep up. I haven't tried bulle, but I'll have to see how my fixed ralph1 car does. Ralph3 (unreleased) is about as fast as the wappucar i have. > I actually saw a funny behaviour from Bulle on Jerez. Bulle is usually very cautious when it comes to overtaking - I have a "middle lane" driver that Bulle has great difficulties passing (the same goes for ralph2, btw). 3 curves in a row is the death knell for most of my cars. They perform like most other cars. I have toyed with ways around this but haven't found a good one yet. It almost makes me want to make a learning algorithm just so that I don't have to figure it out myself. > But about halfway through the 73 laps race, when Bulle was at 5th place or thereabouts, at least one lap down, it suddenly decided to go for it (having fastest qualification time, btw), and started overtaking other cars like a real pro. > And sure enough it won the race! By a large margin. > I've never seen Bulle (or any other car) change behaviour like that. Usually cars tend to slow down as they accumulate damage instead of going faster and overtaking in a wild manner (compared to the usual Bulle behaviour). Ralph1 would do this. ---ralph ------------------------------ Date: Fri, 14 Nov 1997 00:55:03 GMT From: Robert Wilderspin To: "feverish.rars" Subject: Error in Road America? Message-ID: <346ba164.983582@127.0.0.1> When fiddling around with my ex-midget on the RA track, I noticed that it was behaving strangely on one of the straights. Inspecting the track file, I saw that it was not a straight, but two straights, with a *very* short curve between them, called the Monraine Sweep. Fine, I thought. My car had been slowing down for this, so I changed the code to take account of small corners like this by examining radius/length. However, it still didn't fix it for that curve, even after I went through the figures with a pen and paper. So I put a line in my code to printf the radius and length at that point, and what did I find? For some reason, instead of reporting the radius as 101.75, it was 21.75. I edited the track file, changed it to 171.75, and my car reported it as 91.75. Length is being correctly reported as 0.0526. The question is, why is it doing this? As far as I can see, it's only doing it for this curve, but it was enough to throw my car around. It was only because my car is coded so badly that I noticed. :-) rob. ps. I may enter the thing as traffic for the rest of you. It can get about 88-89 on a clear track, but it has no collision avoidance and only the basic pitting stuff that Torben put in. It usually rams the back of something that's going into the pits... -- Anything that beats my car is obviously cheating. I'll get even one day... -------------------------= (send replies to rob@) ------------------------------ Date: Fri, 14 Nov 1997 01:16:58 GMT From: Robert Wilderspin To: "feverish.rars" Subject: Qualifying Lap Times Message-ID: <346ba61c.2191624@127.0.0.1> I don't know about anyone else, but I like to know how my car compares with other people's cars. Comparing it against the standard cars is becoming less useful, as the cars entered for races are usually of higher quality, so I'd like to suggest that someone keeps a record of fastest qualifying laps for each track. A record of "average" lap times would also be handy, because we're not all striving to be the best, just to keep up with the pack. I've gathered some information from the race results (although qualifying times were missing from the last results...?), but I'm sure that I'm out of date with a lot of the data. Does anyone have any views/ideas on this? rob. -- Anything that beats my car is obviously cheating. I'll get even one day... -------------------------= (send replies to rob@) ------------------------------ Date: Fri, 14 Nov 1997 09:30:38 +0100 (MET) From: Maido Remm To: Rars mailing-list Subject: Re: Qualifying Lap Times Message-ID: Content-Type: TEXT/PLAIN; charset=US-ASCII On Fri, 14 Nov 1997, Robert Wilderspin wrote: > > I don't know about anyone else, but I like to know how my car compares > with other people's cars. Comparing it against the standard cars is > becoming less useful, as the cars entered for races are usually of > higher quality, so I'd like to suggest that someone keeps a record of > fastest qualifying laps for each track. A record of "average" lap > times would also be handy, because we're not all striving to be the > best, just to keep up with the pack. > > I've gathered some information from the race results (although > qualifying times were missing from the last results...?), but I'm sure > that I'm out of date with a lot of the data. > > Does anyone have any views/ideas on this? I would like to see much more statistics implemented for the cars and track. One such variable would be track record - which is automatically recorded in track file (together with car name and date). For the average times I would suggest you to download other robots and compile them into your copy of rars. I use WappuCar, Bulle, Jas19, Lisa15, Gryma1, Jammer, Magic, OscCar2, O1, Ralph2, JackBot, Indretti and Burns. They are all good general Bots, coping with most tracks. I can send you a copy of my Apex8 if this is not published on ftp site. Those cars are also in my WWW page, some of them are found under BORS robots. If link to ESTONIA is too slow, try the mirror in Sweden: http://evolution.bmc.uu.se/~maido/rars/rars.htm Maido ------------------------------ Date: Fri, 14 Nov 1997 09:54:55 +0100 (MET) From: Maido Remm To: Rars mailing-list Subject: Re: Error in Road America? Message-ID: Content-Type: TEXT/PLAIN; charset=US-ASCII On Fri, 14 Nov 1997, Robert Wilderspin wrote: > When fiddling around with my ex-midget on the RA track, I noticed that > it was behaving strangely on one of the straights. Inspecting the > track file, I saw that it was not a straight, but two straights, with > a *very* short curve between them, called the Monraine Sweep. > > Fine, I thought. My car had been slowing down for this, so I changed > the code to take account of small corners like this by examining > radius/length. However, it still didn't fix it for that curve, even > after I went through the figures with a pen and paper. > > So I put a line in my code to printf the radius and length at that > point, and what did I find? For some reason, instead of reporting the > radius as 101.75, it was 21.75. I edited the track file, changed it > to 171.75, and my car reported it as 91.75. Length is being correctly > reported as 0.0526. > > The question is, why is it doing this? As far as I can see, it's only > doing it for this curve, but it was enough to throw my car around. > This is because in situation s RARS reports you always the INSIDE radius of the curve, but track data in track file is read into trackin structure, which can be outside or inside radius depending whether it is right or left turn. You have to consider inside radius when making your bot. > > ps. I may enter the thing as traffic for the rest of you. It can get > about 88-89 on a clear track, but it has no collision avoidance and > only the basic pitting stuff that Torben put in. It usually rams the > back of something that's going into the pits... > -- My car wouldn't beat you on speed (88.5), but it hopefully will in collision avoidance :-) Maido ------------------------------ Date: Fri, 14 Nov 1997 08:22:16 -0500 From: Ralph Scott To: Rars mailing-list Subject: Re: Qualifying Lap Times Message-ID: <346C5088.A21@netusa.net> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Maido Remm wrote: > > On Fri, 14 Nov 1997, Robert Wilderspin wrote: > > I've gathered some information from the race results (although > > qualifying times were missing from the last results...?), but I'm sure > > that I'm out of date with a lot of the data. > > > > Does anyone have any views/ideas on this? > > I would like to see much more statistics implemented for the cars and > track. One such variable would be track record - which is automatically > recorded in track file (together with car name and date). What other statistics were you thinking of? What about an improved .out file? ---ralph ------------------------------ Date: Fri, 14 Nov 1997 08:26:47 -0500 From: Ralph Scott To: Rars mailing-list Subject: Re: Error in Road America? Message-ID: <346C5197.423F@netusa.net> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Maido Remm wrote: > > On Fri, 14 Nov 1997, Robert Wilderspin wrote: > > The question is, why is it doing this? As far as I can see, it's only > > doing it for this curve, but it was enough to throw my car around. > > > This is because in situation s RARS reports you always the INSIDE radius > of the curve, but track data in track file is read into trackin > structure, which can be outside or inside radius depending whether it is > right or left turn. You have to consider inside radius when making your > bot. Yes. Trackin is not a good name. This harks back to when all tracks were run counter clockwise. It then stood for track inside wall. It should be track_left_wall. Because it always represents the parameters of the left wall. If it is a right hand turn, the left wall radius would include the track width. If it is a left hand turn, then the radius would be accurate. My older bots would usually fix this first, and then not worry about it later. ---ralph ------------------------------ Date: Fri, 14 Nov 1997 17:31:29 +0100 (MET) From: Maido Remm To: Rars mailing-list Subject: Statistics Message-ID: Content-Type: TEXT/PLAIN; charset=US-ASCII On Fri, 14 Nov 1997, Robert Wilderspin wrote: > > I don't know about anyone else, but I like to know how my car compares > with other people's cars. Comparing it against the standard cars is > becoming less useful, as the cars entered for races are usually of > higher quality, so I'd like to suggest that someone keeps a record of > fastest qualifying laps for each track. A record of "average" lap > times would also be handy, because we're not all striving to be the > best, just to keep up with the pack. > > Does anyone have any views/ideas on this? > More thougths: As a first thing Ralph could include additional column in results 'pit_stops' . It would be interesting to know after race. This is easy to do. In future (next year) I dream about Rars on-line statistics as following: There would be only one 'LEADER BOARD'. Car order will be according to current position on track. By pressing SPACEBAR or PAGEDOWN we could see following info for all cars. LAPS LAPS STARTING COMPLETED LEAD POSITION ___________________________________________ FUEL DAMAGE ___________________________________________ PIT STOPS LAST STOP ON LAP ___________________________________________ SPEED AVERAGE LAST LAP BEST_LAP __________________________________________ BEHIND BEHIND AHEAD LEADER NEXT NEXT Plus track record , record owner and date (kept in track file) together with track length. Instrument panel can stay with tan_g, lat_g, alpha, v, vc Do you think that is too much to dream? This is in my list of works. (somewhere near bottom :-). Any volunteers? Maido -------------------------------- End of rars-d Digest V97 Issue #5 *********************************