3D Robotics

T3-Round 5: Virtual Sparkfun competition!

For the fifth T3 (Trust Time Trial) contest, we've got an easy and fun one: set the best time for a virtual Sparkfun Autonomous Vehicle Competition course. The real Sparkfun contest will be held on April 17th in Boulder, Colorado, and is simply the fastest time for an autonomous vehicle to do one lap around the company's headquarters (shown above). The DIY Drones team won it last year in a time of 35 seconds, and we (and other DIY Drones community members) will be entering this year, too.

Your mission is to program your UAV to fly a pattern around a rectangle the size of the Sparkfun HQ (dimensions shown above). One lap, fastest time wins. The rules don't say anything about takeoff or landing, so you can cross the start line (shown in the picture) at full speed and it doesn't matter what happens after you cross the finish line. Altitude is up to you, but stay below 400ft, as always, and I'd recommend staying above 100ft to avoid the evil trees!

I expect that competition will come down to fractions of a second. Since most GPS loggers only record a position once per second, you'll need to interpolate between timestamps for both the start and finish time. The way to do this is to draw a line from the pointy bottom tip of two adjacent Google Earth pushpin markers. Estimate how far along that line the start or finish lines cross, and apply that fraction to estimate the real time. This illustration gives an example:

Please show these calculations on your KML path image.

BTW, it would be cool to have a model of the Sparkfun HQ to place in Google Earth so you can display your path around a "real" building. Your Chief T3 Judge, Gary Mortimer, has issued a call for a volunteer to make that model, so when one of you rises to the challenge I'll post it here for all to use.

Same submission rules as always. Submit time, autopilot and airframe, and KML track (both screenshot image and raw data) as evidence in the comments below. Video very much appreciated but not required.

Deadline Wed, March 14th, 12:00 midnight PST.

Gentlemen, start your engines!
E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • 3D Robotics
    Sorry, some formating in the GCS output caused those times not to appear. Here they are again:

    Reached Waypoint 4 on Sun Mar 14 17:39:58 GMT-0700 2010 -- Start
    Reached Waypoint 5 on Sun Mar 14 17:40:03 GMT-0700 2010 -- First WP
    Reached Waypoint 6 on Sun Mar 14 17:40:08 GMT-0700 2010 -- Second WP
    Reached Waypoint 7 on Sun Mar 14 17:40:13 GMT-0700 2010 -- Third WP
    Reached Waypoint 8 on Sun Mar 14 17:40:19 GMT-0700 2010 -- Finish Line
  • 3D Robotics
    Okay, I'm not really eligible to win, but Jason Short and I wanted to enter anyway. Here's our run: 21 seconds!

    This image is a composite of the KML track in Google Earth, the waypoints from the ArduPilot config utility and a screen grab of the Sparkfun HQ from Google Maps, roughly scaled to match the Google Earth image.

    Here are the waypoint times, datalogged from Jason's GCS.

    First WP
    Second WP
    Third WP
    Finish Line


    EasyStar with brushless upgrade
    ArduPilot 2.5 RC2 code

    KML is here
  • T3
    I totally agree with Joe. Based on that poor judging, I have lost a lot of faith in the T3 competition.
  • Gary and Chris,
    I guess this round is like an invitation for more/new contestants to come forth, and I sincerely hope it will be a success as such. For my part, the previous round was more in the middle of my interests, and I think it was particurlary interesting because it even attracted pro teams. (Lots of learning). But nevertheless, I will try to post an attempt for this round as well - when the weather improves.

  • T3
    Joe, frankly speaking, this was Christmas round and somehow winterish posts got promoted. Let's look this way: so far the scene was dominated by fast flying crazy EasyStars with thermopiled autopilots, overboosted motors and too much lipos to spare. I think the contest is returning to this equilibrium: crazy flying of overly tight, too fast or otherwise 'unnecessarily, counter-productively difficult' patterns. The real reason for the contests is probably the idea: let's do it, because it can be done. Also we like to humiliate any pro-looking contestants - a limited version of 'open community syndrome'. You can look at the contest as mission readiness test. If you can adjust your Flying Machine to the new settings once per 6 weeks, you are preparing your team for the Real Battle. The more blood during trainings, the less in fight. For example I am preparing my EasyUAV for this round knowing well that in 20km/h winds with slight snow and total overcast my results will be at best funny. If the 1yr old motor finally chokes on water and evaporates its winding - even better, will be fun to dismantle it.
  • 3D Robotics

    That last round required subjective judging, which is bound to be somewhat controversial. I'm sorry that it didn't seem fair, but I wanted to assure you that this round is purely quantitative. If you've got the best time, you'll win. Simple as that.
  • I can't speak for the other entrants involved, but the results from the last round were poor enough that I can't honestly tell my team mates that we were judged fairly. As such I simply can't justify investing further time into T3 competitions; particularly with the AUVSI 2010 competition coming up.
  • 3D Robotics
    Weather's bad here. And we've been head down on 2.5. Now that's out, I should have an entry in the next week or two.
  • Developer
    Weather is bad in the UK. (and problems with radio interference in my plane).
  • Moderator
    Er is this round the first one to be a complete failure or is the weather really rubbish up top or is everybody holding back....er .... were my judging decisions so bad last time???
This reply was deleted.