PDF Print E-mail

MOTMEN UNMANNED AERIAL VEHICLE PROJECT

 

Start date of project 15 May, 2007.

MotMen has been working with several specialists who have helped develop our Unmanned Aerial Vehicle project for high resolution imaging. Work initially began after consultation with Silvertone Electronics (Sydney) under the direction of Bob Young, a veteran of unmanned aerial vehicles in Australia and a real pioneer.

 

8 Nov 2008: Preliminary image taken from the air. Focus set to infinity; ISO 400. Nikon S710, 14.1 megapixel. Altitude approx. 150 m.

 

Dec 2008: Here we demonstrate a 2-axis camera gimbal. The gimbal is housed in a camera pod. which we attached to the underbelly of the airframe. The gimbal keeps the camera pointing to the center of the earth relatively independent of orientation of the aircraft.

 

June 2008: We built our own outboard generator system onto the MVVS engine of our Senior Telemaster. We used a Thunder Tiger outrunner and an o-ring. In a later version we used a timing belt to prevent slippage. The design incorporates a 'battery switching' circuit that allows the UAV computer to draw power from the generator when the prop revs are high enough to generate sufficient power. The system flight tested successfully.

Feb 2009: Battery-generator switching circuit 4-8d.

 

Feb 2009: Close up of the actual board from above circuit diagram (4-8d). Very crude, but effective.

 

Dec 2008: Detail of our camera pod version 5c. Weight: 1.8 kg.

 

March 2009: Pod5c wiring schematics.

 

Then came the vibration dragon. How do we get rid of image blur caused by engine vibration? The solution was to stop and start the engine in flight. In the meantime, we shot some nice aerials using the GoPro.

 

8th May, 2010: GoPro HD camera test. The camera was packed in high-density foam and was carried by a Telemaster with its engine switched off. The video can be viewed in full HD (1920 x 1080, 30 fps).

 

Saturday, 2 July, 2011: We flight tested our starter motor system for the MVVS 26cc engine on our custom built Telemaster. The starter motor worked flawlessly. The next challenge is to add a generator to the back of the crankshaft of the MVVS so that we will be able to maintain electrical power for the imaging and navigation systems on long duration flights.

 

2 July, 2011: The engine easily kicks over from a cold start. No more hand spinning!

 

Autopilot development has progressed. We have completed static ground test. Aerosurfaces behave correctly when orienting the airframe. We are using the arduino system which so far has been workable (http://code.google.com/p/ardupilot/wiki/ArduPilot).

 

22 August, 2011: The 'ArduPilot' (v2_7_1) based on the arduino board. We configured the system as a single mountable unit powered by the servo batteries.

 

9 Sept, 2011: The ground station setup. A USB cable converted to a Data Interface Cable snakes its way from the 10 inch EEEPC AH1000 notebook to the interface port of the side of the airframe. Programming parameters are uploaded to the arduino board using arduino software (http://arduino.cc/en/Main/Software). Note: we had to reverse our servo settings.

 

 

Side view of the Data Interface Port. You can also see part of the ArduPilot module (Xbee aerial, and IR sensors). The version Xbee is: XBee-PRO® XSC, 2.4 Ghz, Extended-range multipoint, 1 mile / 1.6 km, +18 dBm, -100 dBm, Through- hole, 250 Kbps, S1. See data sheet 'chart_xbee_rf_features.pdf'.

 

Sat 17 Sept, 2011: Maiden test flight of autopliot successful. We estimate 500 cycles of engine stop-start per flight with a small fully charged battery.

 

Mon 7 Nov, 2011: Tested the ardupilot 2_7_1 through Oct to early Nov. Having huge problems with altitude hold in AUTO mode. We have focussed on the throttle as we run a petrol engine whereas ardupilot was originally configured for electric engines. It seems that nothing we do solves the problem. Then on Sat 5 Nov 2011, trying to keep the nose down (setting the Pitch Trim down in the config file), going from STABLIZE mode to AUTO resulted in a radical nose up which put too much stress on the wings. The right wing detached and the plane plummeted to Earth. Looks like it's back to square one. We have decided to completely overhaul the code to get to the bottom of this problem.

 

18 Jan, 2012: Completion of rear mounted generator prototype.

Crank shaft adaptor to run Thunder Tiger 750 rpm/V outrunner. The housing has to be totally enclosed to maintain piston compression.

 

24 Sept 2012:

Rebuild Senior Telemaster airframe with strengthened wing struts.

Continue debugging ardupilot. Work up detailed schematic of ardupilot integration.

 

Ardupilot-radio integration schematic 1-0.

 

Friday 14 June, 2013:

After messing around with the ardupilot and getting no where, we decided to shift to the ardumega AMP2.5 (http://code.google.com/p/ardupilot-mega/wiki/home). The initial stated goal was to integrate the thermopile sensors into the APM autopilot code. The reason for this was that we believe that it will offer additional redundancy for stabilization on a high vibration load platform like ours running a petrol engine.

Attopilot XYZ thermopile sensors used in this project. For more information and data sheet see:

http://paparazzi.enac.fr/wiki/Infrared_Sensors

 

22 July, 2013:

After several weeks of hard work, we have achieved our initial goal of integrating the thermopile sensors into the AMP 2.78 firmware code. Tests showed that time-wise the thermopile sensors are comparable to the gyro. There was a slight shift in the data plot possibly due to a) tests being conducted close to our warm bodies (we held the APM module in our hands for testing), or b) being too close to the ground, or c) they simply need calibrating. In the near future we will conduct air tests to tune them to the level of precision needed. Those air tests were conducted on August 10, 2014.

 

22 July, 2013: AMP2.5 gyro versus thermopile raw roll and pitch data.

 

Friday, August 1, 2014:

Bixler 2 build complete. APM 2.6 installed. Thermopile sensors installed and APM 2.78 firmware modified to acquire thermopile sensor voltage data from analog pins 5, 6 and 7. Our LabVIEW ground station is mostly complete, except for a few bells and whistles, it does all the necessary things required for monitoring flight and collecting telemetry. An alpha version of our LabVIEW ground station can be downloaded here MotMen_GS_4.29k.

 

Cradle made from PVC plastic for AMP 2.6 with power module.

 

AMP 2.6 integrated into the Bixler 2 airframe. The cradle assures that the AMP is aligned and level inside the cockpit. A cut was made in front of the cockpit so that the power module XTC connector could be plugged into the battery (which is tucked under the cradle and held in place with Velcro. The XY thermopile sensors can be seen positioned above and aft of the cockpit area. Note that each sensor has a clear and unobstructed view angle. The Z sensor is located in the nose.

 

Flight configuraiton showing 'hood' held closed with a rubber band.

 

Sun, August 10, 2014: Bixler 2 maiden flight. Collected in situ thermopile data.

 

 

The thermopile data parallels the gyro data relatively faithfully. The slightly narrowed range of the data is due to a 10 percent kalman filter applied to the data. Comparing the data collected close to the ground and in flight shows that as the aircraft gets closer to the ground, the thermopile sensors lose their reliability.

 

Sun, Sept 21, 2014: Soldered up a uni-plug, which replaces individual servo leads on the input side of the APM (from RC to APM input BUS) with a single plug. Product description see https://www.mill-max.com/new_products/detail/81. Can be ordered from digikey at http://www.digikey.com/product-detail/en/805-43-024-10-012000/ED8054-24-ND/2552608.

 

Best way to solder the wires is do the middle pin first for each triplet. You also need something to hold the plug at different angles if your fingers aren't nimble enough to hold it and solder it with two hands.

 

12 October, 2014: Screenshot showing the maiden autopilot track of the Bixler2 with APM 2.6 onboard. The thermopile sensors were disconnected during this test, so the 'thermo horizon' is inactive.

12 October, 2014: Youtube video showing a short excerpt of the flight. The Bixler gracefully peels away toward its next weighpoint at the end of the shot.

 

25 April, 2015: Daniel with the 3-polarizing camera navigation system setup mounted on top of the main wing on the Senior Telemaster just before the first test flight.

 

25 April, 2015: Close-up of the three-camera system. Note the judicious use of a PET bottle as a cowling.