Blogs

Moderator

T3 Back to the Future

Posted by Gary Mortimer on May 27, 2015 at 8:00am

3689652774?profile=original

Remember back, way back in autopilot years probably about 20. The year 2009 and the first round of our gentlemens (ladies are allowed just none have ever entered) competition the T3 challenge.

Back then just flying way points was hard and all this gear was expensive. Ardupilots were made at Sparkfun in batches that ran out quick.

3689652827?profile=original

Lets try it again I suspect this time the challenge would really be flying it faster than somebody FPV racing.

Read about the first time here for hints and tips. 

As tech has moved on all aircraft types welcome lets make it not above 50m and 5 laps. All of course autonomous. Extra marks if that is from take off to landing.I will count the time from take off to landing so screaming dives to the start will not help. KMLs as evidence please!

UPDATE: Rovers (and boats) are also allowed! There will be a special rover/boat class, and you only have to beat other vehicles in that class. The course for such ground/water vehicles is 100m x 100m. 

I will accept FPV times for the sake of science and should they be much much quicker it will give the pure FPV crowd something to brag about.

An Easystar (yes before HK copied it) made one lap in less than a minute in 2009 so 5 in 5 minutes should be more than possible. 

As ever the time frame is flexible but why don't we all meet back here on August the 1st and find out who won!

Please add your entries and adventures below. Points are always awarded for silliness, just as long as its safe!

3689652856?profile=original

E-mail me when people leave their comments –
Follow

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

Join diydrones

Comments

  • dr0n3droid August 8, 2015 at 7:50pm
    Well I just converted 30 .bin files to log, created kml files, and they are all blank, with one exception and two that showed me in the middle of the ocean. What am I doing wrong?
  • dr0n3droid August 8, 2015 at 7:21pm
    Looking for the log right now. Wait till you see it. Way off course. I'd never had such issues with APM2.6, only since upgrading to pixhawk.
  • T3
    healthyfatboy August 8, 2015 at 7:19pm

    Thanks! It's definitely not tune all the well as you can see it oscillate on every straight but it still did well. I wish it was light so I could have stuck a Mobius on it but oh well. Maybe another day.

    Good luck with your setup! I know you have had a lot of things to deal with trying to get it to work.

  • dr0n3droid August 8, 2015 at 7:11pm
    Very impressive. I managed to complete a few laps yesterday in the grass, but the results are embarrassing to be honest. I'll post it tonight.
    Not sure why I can't get this thing to track a straight line. It may have to do with the stack of errors.
    Bad compass health, err.compass variance, bad ahrs, pos vert variance. Rover wanders 30-40 meters off course before correcting. I may try again in the morning.

    Mike
  • T3
    healthyfatboy August 8, 2015 at 6:59pm

    Okay, here's my entry. I had to borrow a Pixhawk as I had issues trying to get any logs off my APM for whatever reason. It ran just like my APM once I transferred the right parameters over. It's pretty amazing how seamless it worked going from one AP to another!

    3702065447?profile=original

    My run took 373 seconds to do 5 laps in an empty lot. I had to do it once the place closed but I finally got a good run in last night. It looks like I was hitting 9 ms^-1 on the straights and it just went like it should and was a pretty good run. I was happy with it given how little time I had to mess with tuning but it did a decent enough job. The rover had to slow down considerably at each waypoint so I was getting down to the 1-2 ms^-1 range there but that was mostly out of necessity. There were curbs all around that I didn't want to hit. The actual course is probably ~100x97x100x100 from waypoints 1->2->4->3 but I really couldn't have gone any larger as there is no other place here I could do it so I got as close as I absolutely could without too much risk to the vehicle.

    Being adventurous and having plenty of battery life, I bumped up THR_MAX to 100, adjust the CRUISE_THROTTLE to 12 ms^-1, and adjust a couple other parameters to account for the higher speed and I went at it again. Here's what it looked like.

    3702065604?profile=originalI had to go to manual mode four times on this one. Once right off the bat because I nearly hit the light pole with the large concrete base doing the cross from waypoint 2 to 3. I had to do it again for the same reason going from waypoint 10 to 11. Then I overshot waypoint 17 and went over the curb into the grass and seeing I was 100 m away, I didn't want to take a chance rolling it over on the grass since the Slash wasn't covered on top. I also didn't know what was there as it was dark. Finally, from waypoint 19 to 20, I put it in manual a split second before I hit the concrete light pole at 11 ms^-1. Ouch! Assuming the logs were correct, I hit 80 g's on impact. I should have been standing past the light pole to see where the truck was going but instead was before it and it was too late to recover. Had I not needed to go manual the 4 times and had I not bashed my truck into the light, I would have finished in under 300 seconds. It was quite the nice run most of the time.

    After putting everything back together, I thought I would tempt fate and give it another go. I had some battery life left but knew it would be getting weaker with one more run. I got really close on the last one so why not? Here's the result.

    3702065571?profile=originalI found that near the pole I hit, it went the other way and clipped the curb the first time and I took it out of AUTO. The next time through, it did the same thing but I just let it keep going. Fortunately/unfortunately, my receiver came undone during the transit over the curb and the rover went into HOLD a second later and coasted near the far curb about 40 m away. I found the receiver at the curb, picked everything up, and called it a night given the rover gods saved my bacon by dislodging my receiver before I did any more damage.

    Here are some lessons learned and observations:

    • Set your failsafe properly! Had I told it to continue in AUTO or anything else, it could have been messy. I was a sprinter in high school but even in MANUAL mode, it would have gone full reverse and running >11 ms^-1 would have been hard holding a phone, Bluetooth Bridge, and Taranis.
    • The throttle failsafe timeout default of 5 seconds would have been interesting too. I'm glad I set it to 1 second but even then, it went another 40 meters from me after I physically lost my receiver and during the subsequent coast.
    • HOLD mode should have some form of braking. Maybe full reverse for 1 second and then let off. If you're running a slow mission or in grass, it's not a big deal but on pavement, there needs to be something else so it doesn't coast into oblivion.
    • I also found that I can't set NAVL1_PERIOD using the most recent Tower app. I'm running 5.1.1 Lollipop on a rooted Nexus 6, the latest Tower app version, and a Bluetooth Bridge and every time I would go to adjust NAVL1_PERIOD, it would crash Tower and leave me hanging with no telemetry. Having a telemetry module connected via USB would have definintely sped things up during each app crash!
    • It looks like the average current during the straights was around 10 A or so but during braking, it would jump from 60 to 120 A on the first run. The data showed up to 180 A spikes on the faster run! I don't know how correct that is since this the power module only rated to 90 A but yikes!


    I'm glad I got some good runs in and a complete mission end to end in AUTO. It felt really gratifying knowing I completed a T3 challenge as I have only read about it in the past. Thanks again for extending just a little to let me get this finished!

  • T3
    healthyfatboy August 7, 2015 at 4:38am

    Yup!

    I need to figure out how to get the proper logs/track from my rover as what I did successfully last night didn't seem to turn out so well when I looked in MP. Hopefully I can get that figured out quickly as I only have a couple of nights left before the rain comes in.

  • Rob_Lefebvre August 7, 2015 at 4:16am

    So the competition is still open?  Weather looks good for this evening.

  • dr0n3droid August 6, 2015 at 7:48pm

    The esc is toasted.  Upon inspection I couldn't believe this didn't happen sooner.  The bullet connector for one phase was completely exposed on one side, while the other two have many nicks in the shrink. 3702064427?profile=original

    3702064283?profile=original

    And the replacement, for comparison.3702064498?profile=originalI think I'll give up on the high speed attempts.  Right now I'm just hoping to avoid a DNF.

    Mike

  • Felipe Rabat August 6, 2015 at 4:54pm

    Well, I really wanted to participate on this one, but unfortunately, I took off in my Y6 with a pack that was not fully charged to film my roof, and I had a "premature landing" on the roof, and the slope of the roof cause a "secondary landing" on the grass below, which cost me 2 motors and a motor mount.  I have the 2 motors, but don't have any spare mounts.  Already contacted 3DR and they are checking inventory for me, but unfortunately, I am grounded until I get the new motor mount...

  • dr0n3droid August 6, 2015 at 4:31pm

    Hey Rob,

    I've been running the same motor esc setup for a few years now without issue.  Looking over a few logs, I could not find any current spikes over 55-60 amps @ full throttle. The other night, I didn't exceed 40% throttle. Made a 2-3 min run, walked back to the car, sat for a few mins while I made a phone call, then when I tried to make another run, the pixhawk browned out when I touched the throttle. After reboot, the same thing happens. steering input is fine, but throttle causes brown out.  I let it cycle a few times before the PM gave out.  The wetness is the only thing out of the ordinary, which brought me to that conclusion.  That, and the fact that there was grass jammed in there.

    However, it's on the bench right now, I've installed a new PM, but the problem persists.  I'm looking over all wiring and connectors for a short, then I'll swap out the esc. It's not good, whatever it is.

    Mike

This reply was deleted.