autopilot: UAV command and control / News: Recent posts

Aaron and I did autonomous flight tests this afternoon. We were able to give control of the pitch axis to the software and tune the control parameters. We then did the same for the roll axis and then let it take over both of them.

IT WORKS! Yeah! One year after starting the project, we're finally flying under computer control. It's time to party, then revamp the website.

The helicopter simulator aerodynamisc have been improved by Aaron Kahn to use blade element in combination with momentum theory, giving it much more accurate hovering dynamics.

Details, screen shots, movies and source can be found here:

http://autopilot.sourceforge.net/sim.html

autopilot is a command and control system for helicopters (and can be extended for fixed wing aircraft). It provides a three-axis EFIS and moving map on the ground station. The unmanned aerial robot has instrumentation for attitude, engine and position.

Release 2.0 adds support for the
custom PCBs and the Atmel AVR Mega163 microcontroller. The code also underwent some major reorganization.

Download it!... read more

I've written up details on building the N1 (engine) and N2 (rotor) tachometers for our large prototype:

http://autopilot.sourceforge.net/tach.html

Includes pictures of the installation on the Kyosho Concept 60.

The Rev 1.0 IMU board has been sent to ExpressPCB for production. You can see the board layout in onboard/pcb/autopilot.pcb tag "rev-1_0".

The boards will be ready Tuesday the 8th of January. Gyros, Accelerometers, and microcontrollers still need to be ordered.

Please review the combined circuit board for the IMU and servo controller:

http://autopilot.sf.net/pcb.html

I plan to send it off for fabrication soon and would like to have more eyeballs look it over. If you want to get in on the group order please send mail to the list.

After nearly a month of waiting, the parts arrived to repair the large prototype. Anthrax and other problems have delayed the mail getting into DC.

We test flew it yesterday at Lee's and had several successful manually controlled flights on the training skids. With the gyro correctly wired the flights were fairly tame.

The IMU board has been merged with the servo controller board, saving quite a bit of weight, power and cost. Details are available:

http://autopilot.sf.net/pcb.html

Please review the circuit and make any suggestions. We plan to make a run of the board in a week.

The ADXL202 output is now being fed into the groundstation's EFIS display. We're using the PWM output rather than the analog voltage on the sensor; Analog recommends this method.

First flight with it will be when ever the new blades arrive.

The large prototype was "flown" for roughly one minute this morning before a horrible crash took out many components. We're waiting on main blades, tail rotor blades, linkages and tail boom. Maybe next week we'll be up and running again.

It turns out that all three crashes have been due to a reversed gyro. That has been fixed, but should have been caught much earlier.

The small prototype maingear has been damaged due to a bad mesh with the metal drive gear. Several teeth were destroyed and the resulting shock load on the tail rotor drive gear caused it to strip off the drive wire.

Parts have been ordered. They should arrive in a week or so.

Due to a tail-rotor pitch control slider failure, the second prototype is out of commission. The slider unthreaded and caused a loss of the anti-torque rotor effectiveness. The resulting spin impacted the tail rotor blades and broke one of them.

New parts are on order. The pitch control slider has been repaired and rebalanced.

Release 1.4 and 1.5 added support for PWM gyros and analog voltage ones. The MG100 is fully supported in a temperature calibrated mode.

The rate integration code is failing to produce accurate orientation data, so it will be augmented with a two-axis accelerometer. These should arrive in a few days.

The 1.3 release adds support for the AVR microntroller and offloads all servo processing tasks to it. This eliminates the need for the MiniSSC and gives the onboard system the ability to sample gyro and servo inputs as well.

We were able to perform a demo flight for a small audience last night. The prototype hardware lifted off, hovered and then crashed into a curb. Just like before.

Except this time we had the video camera for attitude reference and the controls were being passed through the laptop. Once an AHRS is located, the autopilot should be able to take over.

Since the TINI board runs only Java, I've started porting the onboard code to that language. Despite my serious misgivings about it, I might add.

I'll put together a code release for anyone who might want to try it out. The code base should remain compatible between the Java and Perl versions for a while. The groundstation will remain in Perl/Tk.

Not that anyone cares, but the website is now w3c compliant HTML 4.0 transitional.

I've repaired the tailboom with a wooden dowl at the weak point. It appears to be considerably stronger, as well as a little bit heavier. That will help counter the forward CG of the camera.

Release 1.1 now has working force-release trim!
Hit the trigger to "latch" the trim for the cyclic and the anti-torque rotor. It makes trimming out different forces much easier than using the "hat switch" trim that was in place before.

For fine tuning, the hat switch still works.

The Mini-SSC servo driver has been interfaced and is now controlling the flight control surfaces. Unfortunately the repair parts have not yet arrived, so it just moves them on the ground.

I'm going to start on the embedded system next.

Tailboom, gearbox and fin parts have been ordered. After last week's awful crash, the prototype has been sitting. The replacements are supposed to arrive in a few days.