Testing Atto and Ardu at the field--success and near disaster

I've got twin EasyStars, one with AttoPilot and one with ArduPilot. I had them both out today to begin the process of benchmarking the two autopilots against each other. Two interesting things happened, one scary, the other exhilerating:

1)
It's quite windy at our field these days and I've been having trouble with the the current ArduPilot navigation algorithms, so I set out to try various PID settings. I put the EZ in autonomous mode and just started watching it hunt for waypoints, usually ineffectually, while keeping an eye on the ArduStation to diagnose the problem. I kept the plane at a pretty high altitude, and after looking at the screen a bit too long I looked up for the plane and it was gone!

I could still hear the motor, but it was nowhere in sight. I switched it into RTL mode and scanned the skies. Nothing! And the motor got fainter and fainter. Argh! Panicked, I booted up the laptop, switched the Xbee to the Sparkfun adapter and plugged it in, opening up a EVDO wireless Internet connection and loading the ArduStation software to see if I could find the plane on the moving map.

At this point, the plane had been gone for at least ten minutes, and I was sure it had gone out to sea or to another county, but when I called up the moving map, I was delighted to see that it was still in the air, still moving and apparently right overhead! My eight-year-old son started scanning the sky right overhead and with his great eyes he saw it, at a tremendous altitude. Eventually I could see it, too, and I switched to manual, cut the throttle and put it in a dive. A minute later it was landing right in front of us. Whew!

I'm not sure why altitude hold didn't work, but RTL sure did. If we hadn't had the wireless telemetry, I would have assumed the plane was gone and driven off. Many minutes later, it would have gently returned to earth right where we'd been standing. Yay Xbee! BTW, while replaying the video from the flight, I could hear that the RC system was losing its signal really often, with the ESC rebooting and all sorts of scary stuff. I could also hear the 1Hz buzz of interference from the Xbee module. I'm wondering if the Xbee or RF noise from ArduPilot is messing up my RC system. More testing required...

2) I also tested AttoPilot in the other EasyStar. Yesterday's trials had shown that the default gains were way to high for the EZ, so I'd brought them down from the default 10 to 7. Putting it in the air, it was clear that was too high, too, so I brought them down to 5. That turned out to be perfect, and I let it fly circles overhead.

It was eye-opening. Dean has written navigation algorithms that put ours to shame, The EZ didn't just sorta-circle or do random figure eights overhead, it sped around in perfect 100m radius circles, like a protractor. Wow. We've got a lot of work to do to equal that. Next weekend I'm going to have it do the T3 contest, and I think you'll be impressed.

BTW, here's the rest of the UAV fleet that I'll be benchmarking over the next month or two:

1) EasyStar with ArduPilot 2.3 (EM406)
2) Superstar with ArduPilot 2.3 (uBlox 5)
3) EasyStar with AttoPilot 1.8
4) EasyGlider with ArduPilot 2.3 (uBlox 5)
5) Funjet with Paparazzi Tiny 2.1
6) Superstar with UAV DevBoard 2

I've retired the old PicoPilots and my first gen Lego Mindstorms autopilots. I'm also sticking to park flier aircraft, so nothing that I can't hand launch and land in grass for now (that means no Predators or big trainers)
I've

Views: 218

Comment by I Heart Robotics on August 16, 2009 at 9:22pm
It is great to hear that the RTL worked.

I have seen +/-1Volt swings on unshielded wires on an oscilloscope from a nearby transmitter.
A little heavy duty aluminum foil can go a long way towards reducing RF noise.

I have only ever had my ESC reboot when pulling too much current from the batteries.
Do you have a way of detecting brown outs?
Comment by Torin Segstro on August 16, 2009 at 9:24pm
Perhaps it is time to use a barometric altitude altimeter on the Ardupilot instead of relying on GPS? Glad to hear the XBee saved the day.
Comment by Jack Crossfire on August 16, 2009 at 9:59pm
The GPS plots would be nice to see. Interesting that a 3 channel easystar is enough to pull off anything in the wind, with the right algorithm. Wonder if Ardupilot had the full pitot tube, barometer, & 5Hz GPS.
Comment by Angus Peart on August 16, 2009 at 11:30pm
Do you use a 2.4GHz RC system?

3D Robotics
Comment by Chris Anderson on August 16, 2009 at 11:54pm
I am using a 2.4Ghz RC system and 900Mhz Xbee. EM406 GPS. Never had trouble with altitude control before, but note that I've got the throttle on manual.

I think the problem is RF interference from the ArduPilot serial port.
Comment by Jonas on August 17, 2009 at 3:05am
Chris I think you are right! Maybe it is a idea to make a little cage of faraday around the system.
Comment by Peter Meister on August 17, 2009 at 8:54am
Chris,

Dejavu...Yesterday, I did my big flight - 10 miles round trip. 3/4 of the way through. My lipo batter started acting up. I disabled the ardupilot and went manual. I had the video goggles on so I could see and the long range UHF for radio. But without sufficient power remaining and at only 350 ft AGL. I had no choice but to chose a spot in the desert and put her down. Luckily for me, I also use the XBEE. And with the lovely google earth and the accurate positioning give or take a few meters. I was able to drive out in a desert mind you, and locate her in less then 10 minutes.

On that note, the ardupilot flew the route staight as a been. I mean she flew so perfect it was sick. If my lipo had not failed, which by the way she did. I guess after 520 cycles its ok to die, shes dead, 2 cells at 0.0 volts now. But The AP flew perfectly. Not a single mistake. She flys on rails now...and everything I do a mission it just works...Its awesome....

Developer
Comment by Jordi Muñoz on August 17, 2009 at 10:37am
Ok i got it! I will work more in the navigation algorithm, right now is very simple and straight forward. In the other hand i never read anything about control. I should open the books now to catch some ideas.

Chris remember that easystar uses throttle and elevator (tiny bit) to hold altitude if you control throttle manually (and use enough power) the tiny elevator that is trying to pull it down will not fight back the throttle vs airspeed vs big foam wings.

Also the ardustation has a menu that tell you the direction of you aircraft (respect north) and the altitude angle respect you.

Also the "gone phenomena" happen very often! Thats why i added orange strips to the wings and today i just added very hing intensity 12 volts LED's in the wings (red and blue) to spot it easier.

Developer
Comment by Mark Colwell on August 17, 2009 at 11:49am
Peter, Sorry about your lipo's death, it had a good life ,but remember a new replacement will be better, weigh less and have higher discharge current, longer life too! Long flights are hard on batteries, don't push discharge too long or short life will result, Do you monitor main battery temp over telemetry? It will increase fast over 120ø when over discharged. Great results, I hope to make my maiden flight today.
Comment by Donal Botkin on August 17, 2009 at 6:00pm
Hi Chris et al,
We've come a long way from the days of Mindstorms NXT flying at NAS Alameda! Great work and kudos to you and Jordi. Has anyone written a routine to capture waypoints directly from the Ardupilot? I want to "walk the field" and "push a button" to capture the GPS info for the limits of my flying area and store in eeprom. I'll write one if none exists, but don't want to reinvent the wheel.
Thanks,
Donal

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service