Baker Lake flight

[Mark Hanson]

In June we flew our amatuer UAV, EasyLander over Baker Lake again.
This is the fourth year now that we've tested PICOPILOT in this plane, flying the 7 mile long Baker Lake autonomously (hands off). EasyLander is powered by a AXI2208/26 electric motor and a 2500mAh 2S LiPo.

This year, 3 waypoints were created on Google Earth and exported as KML files which were, in turn imported by the PICOPILOT Editor using the new KML feature. PICOPILOT automatically saves the launch location as the last waypoint. Like previous BakerLake flights we followed along in our boat so the plane was always in visual range. Surprisingly, the little (23 oz) plane has to be trimmed for slow flight because our boat will only go 25 mph.

After hand launching from the boat, PICOPILOT first flew about 4 miles to the first waypoint (mid lake) and turned about 30 degrees towards the end of the lake, 3 miles away. At the end of the lake it turned 180 degrees and headed back towards waypoint 3 (mid lake).

With steep mountains all around and 10,000ft Mt.Baker anly a few miles to the west, a horizon-sensing autopilot would never be able to keep the plane right side up in this terrain. PICOPILOT's built-in wingleveler has no problems.

After about 30 minutes, and about 1 mile from the starting point, it became obvious that the battery was used up and the plane began desending. We've caught the plane before from the boat but this time there wasn't enough time to line up for a catch so it landed in the water. We turned the boat around and soon fished the plane out of the lake. The plane was only in the water for a minute so the wings kept the plane afloat and the electronics didn't even get wet. I guess its time to buy some new LiPos !

[Disciple4123]

Neat :)

[zlite]

In case it's not clear, Mark Hanson works for UNAV, maker of the PicoPilot

[Mark Hanson]

Here is a view of Baker Lake from a nearby ridge. That's 10,000ft Mt.Baker
in the background.

[dmgoedde]

message removed

[zik]

message removed

[zlite]

Hey Dean, that was a little uncool taking over someone else's thread to promote your own product.

You mean the thread that they started to promote their own product (without disclosing that they were part of the company)?

[zik]

message removed

[dmgoedde]

message removed

[Disciple4123]

May we not spoil acheivement with paranoia, whether warranted or not...
I can see what is going on here, certainly not the most suprising or constructive phenomenon I have ever seen.
Just gradual corrosion of an otherwise productive entity(ies).
Let's just all play nice together, ok? Ditto.

Matt 5:39 - 5:44


.

[dmgoedde]

message removed

[pthompson24]

Now this thread is funny!

Kind of like the meeting in the White House today!

:)

[Chucky_D_Gr8]

Wow, This thread kind of feels like some document from the govt with all the posts covered up (either removed or deleted.)

Back on track.

Mark,
Glad to hear that the electronics stayed dry. I noticed that you have only set 3 way points for this flight.

[Chucky_D_Gr8]

I looked up a map of Baker Lake and I'm kind of impressed at the distance between the waypoints. I can see why you only used 3 points to plot this flight. Looks like approx 3.5-4 miles between each point. Is that really true?

-Chucky_D_Gr8

[Mark Hanson]

That's correct, you only need a waypoint when you want to change course. The autopilot doesn't care what the distance between them is. I rarely use more than 4 waypoints in my "real-world" missions.

Many UAV newcomers forget to account for the turn radius. Airplanes need room to turn and autopilots make relatively slow turns compared to what most RC pilots are used to. In fact most "Pro" autopilots have a max. bank angle to 45 deg.

Example: A plane flying at 55mph and a 20 deg bank angle will have a turn radius of 1/10 mile. That's nearly 1/4 mile to turn around ( 180deg ).

Also keep in mind that FAA rules for "Amatuer UAVs" require you to stay in visual range so you really don't have room to use many waypoints.

[Chucky_D_Gr8]

So if one doesn't have room to use a lot of waypoints, then why do some autopilots have hundreds of waypoints?

[Mark Hanson]

I often use a handheld GPS to create waypoints on location. While that might seem "old tech", it actually has several advantages over graphical methods. The basic PICOPILOT Waypoint Editor was designed for manual data entry (using a handheld GPS).

I like to create waypoints next to physical objects at the flying site. That way it's easy to visualize where the waypoints are. Buildings, trees,vehicles, virtually any object on location can be used as a visual reference for a waypoint.

Graphical waypoint creation methods like Google Earth (or Virtual Earth) actually have several limitations.

creation - you can't create waypoints in the field with Google Earth ( unless you have a WiFi connection for your laptop).

resolution - Google Earth image resolution varies widely by region.The highest resolutions are near major cities, unfortunately those are locations you don't want to be flying your UAV. In some rural areas, resolution is so poor that the maps are useless for waypoint creation.

currency - most map images on Google Earth are over a year old.

accuracy - Google Earth image coordinates are not always very accurate. We've done tests where we compared GE coordinates of a physical object ( in the city) to the readout on our handheld and found discrepancies up to 100 ft. Comparisons for objects up in the mountains were even worse, in one case we saw a discrepancy of 200 ft. !

When figuring the position accuracy of a waypoint, don't forget to factor in the accuracy of the GPS receiver itself. Most modern ( 20 chan ) GPS receivers spec their position accuracy at: 5m to 25m ( 16ft to 82ft ).

Waypoint creation using a handheld GPS is still the most accurate and versatile method.

FYI: PICOPILOT uses the latest SiRF-III ( 20 chan ) GPS receiver.

[Mark Hanson]

So if one doesn't have room to use a lot of waypoints, then why do some autopilots have hundreds of waypoints?

.........marketing hype !

[Chucky_D_Gr8]

Originally Posted by Mark Hanson
With steep mountains all around and 10,000ft Mt.Baker anly a few miles to the west, a horizon-sensing autopilot would never be able to keep the plane right side up in this terrain.
What kind of wing leveler does PICOPILOT use ?

[Mark Hanson]

PICOPILOT uses a turn-rate wing leveler ( TRL ).

TRLs have been around for about 50 years, mostly used in general aviation.
Older general aviation autopilots usually use a pick-off from the turn coordinator to control bank angle.

TRLs have their strengths and weaknesses. They are relatively slow and are limited to bank angles of 45 deg.. They are very reliable though and aren't affected by terrain or weather conditions.

The first use of a TRL in the RC hobby market was BTA's AS-series autopilot (attitude controller). Introduced in the early 1990's, the BTA AS- used a modified mechanical gyro as a heading rate sensor. Although it was marketed as a RC training aid, the AS- wasn't very good at unusual attitude recovery or the "turn and burn" flying style of most RC'rs. Word on the street was that BTA sold a lot more of their AS- units for low cost UAVs and target drones. TRLs are very well suited to slow, stable airframes like UAVs.

In 1998 Aerosonde's "Laima" UAV crossed the Atlantic autonomously and set a world record. The flight took 27 hours and flew over 3000 km. The plane flew through several rain storms and at night. Word on the street was that the autopilot used a TRL and the plane did not have ailerons.

PICOPILOT's attitude control works very much like BTA's AS- and is well suited for simple, low cost UAV airframes.
Of course PICOPILOT also has waypoint navigation capability and uses a 100% electronic design making it much smaller than the AS- units.

[Chucky_D_Gr8]

Why did you say a horizon-sensing autopilot won't work in the mountains ?
How exactly do they work?

[Mark Hanson]

Sensing the horizon to keep the wings level is a clever idea but its pretty simple to understand. Basically they use two pairs of sensors to detect the horizon. One pair looks at the horizon off the wing tips and the other pair looks at the horizon off the nose and tail. If the left wing sensor sees more light than the right wing then the plane is in a right bank (sky = light, ground = dark).

Graupner was the first company (that I know of) to offer a horizon sensor leveler to the hobby market in the early 1990's. Graupner sold them for glider control. They worked pretty good but only on clear days with flat terrain otherwise they could lose control. The early Graupner units used photo-cells (visible light). Later evolutions used sensors sensitive to other wavelengths.

In the mid 1990's, Futaba introduced their PA-1 which was functionally identical to the Graupner units. Futaba introduced their PA-2 in the late 1990's with improved control functions and more selective sensors.
The Futaba PA series was popular in the RC hobby as a training aid because it had the quick response neccessary to keep the RC beginner out of trouble. Even the PA's were not 100% relaible, and would lose horizon reference in certain weather or terrain conditions, but good enough for hobby use.

Around 2000, the FMA CO-PILOT was introduced to the RC hobby market. Similar the PA-2, the CO-PILOT uses IR sensors instead of visible light sensors. The CO-PILOT's IR sensors have improved reliability but it can still get confused by terrain and weather conditions.... good enough for hobby use.


All of the "pro" autopilots today use an IMU for attitude control.

[dmgoedde]

Why did you say a horizon-sensing autopilot won't work in the mountains ?
How exactly do they work?Chucky - I see you are a *very* new member to RCGroups, but have only posted questions on this thread. Welcome. It seems you have a lot of basic questions on low cost autopilots. I suggest (and would expect) that you engage in a 2-way horizon sensing discussion with Paparazzi group members (like AntonK) on other threads.

[small_rcer]

Chucky;

I apologize. I was harsh and my remarks were inappropriate for a public forum. I have removed them.

Small_rcer

[dmgoedde]

SmallRCer: I have done several flights myself where the plane came down due to flight being too long, and propulsion pack ran out. The issue here is not fault of the autopilot or operator.. all were working in good faith.

The plane hit the WPs, the guys followed it by boat, it was a long flight with only 3 waypoints. A success in my book.

[Chucky_D_Gr8]

I may be new to this forum but I've been an RCer for 12 years and I've been doing RC photography for the last 5 years. I've read most of the threads on this site but this is the only one that I found that's talking about actual autopilot flying. I've been wanting to get an autopilot for a while but it has to be a proven design and no one else seems to have one. I can see several disadvantages associated with a horizon sensing type and too many things that can interfere with it. I am considering getting a picopilot-na.

small_rcer,
As stated in the beginning of this thread, Mark has made this run several times. and his plane (not the autopilot) had mechanical failure, and at the end of the run no less. I am in agreement with dmgoedde that this flight was a success, at 14 miles no less.