The TORCS Racing Board
Username: Password: Remember Me?
Lost Password Register
Author: timfoden | Created: 2017-04-25 07:33:08
Subject: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
Now that its time to run the race I realise that I supplied an incorrect mouse_2017.xml file.

Its contents should be like this:

<section name="0">
<attstr name="name" val="Mouse 2017 1"></attstr>
<attstr name="desc" val="1"></attstr>
<attstr name="team" val="Rat Race"></attstr>
<attstr name="author" val="Tim Foden"></attstr>
<attstr name="car name" val="car5-trb1"></attstr>
<attnum name="race number" val="7"></attnum>
<attnum name="red" val="1.0"></attnum>
<attnum name="green" val="0.75"></attnum>
<attnum name="blue" val="0.2"></attnum>
</section>

<section name="1">
<attstr name="name" val="Mouse 2017 2"></attstr>
<attstr name="desc" val="2"></attstr>
<attstr name="team" val="Rat Race"></attstr>
<attstr name="author" val="Tim Foden"></attstr>
<attstr name="car name" val="car5-trb1"></attstr>
<attnum name="race number" val="8"></attnum>
<attnum name="red" val="1.0"></attnum>
<attnum name="green" val="0.75"></attnum>
<attnum name="blue" val="0.2"></attnum>
</section>
Last Edited: 2017-04-25 20:47:14 by timfoden
    Author: timfoden | Created: 2017-04-25 19:11:42
    Subject: Re: 2017 Test Race 4 -- Errors found so far
    For fa1_2017, it seems that due to the way the cars are selected in the endrace.xml file the name supplied for the driver in the fa1_2017.cpp file needs to match with the one from the team registration, so it must be "fa1 driver 1" and "fa1 driver 2".

    The bit of code to change is:

    sprintf(buffer, "fa1 %d", i+1);

    to

    sprintf(buffer, "fa1 driver %d", i+1);

    Cheers, Tim.
    Last Edited: 2017-04-25 19:11:42 by timfoden
      Author: fa1 | Created: 2017-04-26 23:26:10
      Subject: Re: 2017 Test Race 4 -- Errors found so far
      Thanks for raising this, will be corrected!
      Last Edited: 2017-04-26 23:26:10 by fa1
    Author: timfoden | Created: 2017-04-25 20:50:36
    Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
    On attempting to submit the result it became apparent that the result file had incorrect driver index values in the Results/Wheel 1/Drivers section. It had <attnum name="index" val="0"/> for both members of each team. I corrected them so the 2nd driver had val="1" and the result was accepted. I assume this is because the drivers were selected by name in the endrace.xml file and the code doesn't fix up the indexes somewhere.
    Last Edited: 2017-04-25 20:50:36 by timfoden
      Author: wdbee | Created: 2017-04-26 17:28:33
      Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
      Hi Tim,

      this is a bug of the current TORCS Version. I reported it to Bernhard and he wrote a post to the Forum how to handle it until it is fixed.

      See here: Racers read!] Endrace result file cannot be submitted under certain conditions

      Welcome back!

      Wolf-Dieter
      Last Edited: 2017-04-26 17:29:30 by wdbee
        Author: timfoden | Created: 2017-04-26 22:10:15
        Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
        Hi Wolf-Dieter,

        Thanks, it's nice to be back... I'm looking forward to the races but I'm not expecting to do particularly well... the robots appear to have advanced quite a lot while I've been gone! :)

        In regard to the bug... you're quite right. I had previously read that post, but I forgot all about it when I was submitting the race. I'll remember to use the workaround in the future.

        Hey, are we allowed to talk about what happened in test races before the end of the submission week -- I thought this one was quite interesting as well as amusing! :)

        Cheers, Tim.
        Last Edited: 2017-04-26 22:12:16 by timfoden
          Author: firechief | Created: 2017-04-27 04:27:52
          Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
          > Hey, are we allowed to talk about what happened in test races before the end of the submission week -- I thought this one was quite interesting as well as amusing! :)

          Do tell! Its a test race, I'm sure no-one will mind.

          Btw the quality of the main-event races has been extraordinarily high over the last couple of years, I'm expecting this year to be very entertaining to watch. From my own point of view I haven't had much chance to work on my robots for a couple of months, having joined a game-development team (http://deadhold.com) which is keeping me very busy indeed. I'm still determined to find time to compete this year though, I wouldn't miss it for the world :)
          Last Edited: 2017-04-27 11:08:13 by firechief
            Author: dummy | Created: 2017-04-27 10:44:17
            Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
            >I haven't had much chance to work

            Wow, there is hope that someone else could win a race as well this season :)
            I agree, it will be very entertaining this year anyway.
            Last Edited: 2017-04-27 10:44:17 by dummy
            Author: timfoden | Created: 2017-04-27 11:24:18
            Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
            >game-development ... keeping me very busy indeed

            I'm glad you've found some work, I know you were worried about that. Hopefully you'll still find some time to run your robots too.

            >Do tell! Its a test race, ...

            OK, well, I thought the race was interesting for a couple of reasons:

            1. Both of the other teams used car6, and both qualified faster
            than the Mice (using car5).
            2. In the race Fabian's fa1 cars were actually fastest by about a
            second a lap, and it looked to me like the racing lines could
            still be improved to make them faster still. Well done Fabian.

            My amusement was because of something the Mice did. When it came time for the fa1 to cars to pit, they both went in at once, with the consequence that the second one accumulated too much damage and was out of the race. On watching this I realised that the Mice were very likely to do exactly the same thing, so I paid attention when it came time for them to pit too. Sure enough both of them headed to the pits together and I thought I knew what was going to happen, but I was wrong... they both drive right past the pit they should have used (pit 3), and stopped instead at pits 5 and 6, doh! (where of course they failed to pit successfully.) They did the same again next lap and then finally ran out of fuel. On seeing this behaviour I instantly understood what I've done to make this happen -- and that I had planned test it -- but I forgot to do so before submitting.

            Later in the race the remaining fa1 car started behaving erratically after a pit stop and never recovered to normal running. It also failed to finish the race after accumulating too much damage.

            The wdbees worked flawlessly, and of course went on to win.
            Last Edited: 2017-04-27 11:26:19 by timfoden
          Author: wdbee | Created: 2017-04-27 17:14:41
          Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
          Hi Tim,

          being in mail contact with Fabian I used a slightly different version of his bot. He did not make it in time to upload it, I was out of office and did not answer the questions fast enough.

          So the "pit sharing" is not contained in the uploaded version but was implemented in the version I used instead.

          But all the other issues I saw as well. Good to know that it is not a Windows/Linux issue ;).

          Something I hate with the new TORCS version is that the cars do not start at the same position while qualifying. This makes qualifying depending on the starting grid (and the ranking in the starter list). The differences in lap times of both drivers can be over 0.5 seconds!

          Bernhard told me that it is same as with earlier versions but using the tyre model makes it visible.

          I did a quick race with the cars and set the mice to start in front. Without pit sharing they where able to do a pit stop. I did a lot of work to reduce the fuel consume without getting too slow but compared to your car5 or a car8 the car6 is a "Schluckspecht". Bernhard adjusted the tank volume to 92 instead of 91 L for the car6 this year, but the limit for the initial fuel was not increased! So this 1 L more does not really help.

          My drivers never got a chance to overtake, so the mice did win the quick race.

          See you at forza

          Wolf-Dieter






          Last Edited: 2017-04-27 17:14:41 by wdbee
            Author: firechief | Created: 2017-04-27 17:53:35
            Subject: Re: 2017 Test Race 4 -- Errors found in Mouse, Fa1 and race result file
            > Something I hate with the new TORCS version is that the cars do not start at the same position while qualifying. This makes qualifying depending on the starting grid (and the ranking in the starter list). The differences in lap times of both drivers can be over 0.5 seconds!

            Wow, I did not know that, but it does explain some variance I saw last time I was testing. Thanks for pointing it out!
            Last Edited: 2017-04-27 17:53:35 by firechief