Thread Tools
Apr 29, 2006, 08:43 PM
Closed Account
Thread OP
Discussion

Sr. Telemaster + PicoPilot


UAVers,

My name is Jeff Parisse and I own a special effects company in Southern California. We've started a small special project to explore autonomous GPS navigation for a variety of reasons and I'd like to share our results and experiences here.

First some photos showing the guts. GPS58 has six test flights to mechanically trim flight surfaces and balance and next week we will take her out to the dry lakebeds for some PICOPILOT testing.

I'll post specs and details over the next few days... I look forward to sharing with and learning from you guys.

Jeff W. Parisse
www.teslacoil.com
Sign up now
to remove ads between posts
Apr 30, 2006, 12:37 AM
Professor of Wood
kd7ost's Avatar
Thanks for the pics Jeff,

Is that the GPS unit velcro'd to the wall in the first picture?
And the Telemaster must be Electric powered with those big electrical connectors is that right?

Dan
Apr 30, 2006, 11:37 AM
Closed Account
Thread OP
Dan,

Yes, that is a Garmin GSP18-LVC but it is shown in it's storage position. Normally the Garmin unit gets attached to the top center of the wing.

The GPS58 is electric. I'll have a full spec sheet probably today. So far, I get about 30 minutes in the air. I can parallel another pack for twice the time.

Here's another round of photos:

Jeff
Last edited by workshop; Apr 30, 2006 at 02:56 PM.
May 01, 2006, 12:58 PM
Registered User
icebear's Avatar
That's a beautiful ship Jeff! Nice arrangement of the electronics too!

Looking forward to seeing the specs and reports on the test Tuesday!

/Bjorn
May 01, 2006, 01:44 PM
Closed Account
Thread OP

GPS58 Specs...


Wow Icebear... Thank you very much...

Today's Flight Test 7 went off fine. WP32 return mode works. The new prop and motor battery give better performance (don't know the trade-off yet) and the extra weight of the RF shielded enclosure and NiMH batteries (I was using switching PSs) didn't affect flight as far as I could tell. Tomorrow the vast Southern California desert!!

Specs:
Airframe: Senior Telemaster (Hobby Lobby)
Wing Span: 93"
Wing Area: 1330 sq."
AUW w/o Camera: 200.6 oz.
Wing Loading: 21.7 oz./sq.'

Motor: AXI 4130/16
Prop: 16" x 10" APC-E
Battery: TP4200 6s2p
ESC: Jeti 77-O

Receiver: JR 770s (SPCM) Ch. 58
GPS Receiver: Garmin GPS18LVC
Autopilot: PICOPILOT NAV2, ALTE
Rudder Gyro: GWS PG-03
5v PS: 1450mAh, 4 Cell, NiMH
6v PS: 1650mAh, 5 Cell, NiMH
Servos: HS-645MG
Switch Harness: DPDT (both batteries at once)

Jeff Parisse
www.teslacoil.com
May 01, 2006, 04:06 PM
Registered User
icebear's Avatar
Jeff,

Thanks - it'll be interesting to hear how your desert trip goes!
I couldn't do more testing today due to the weather, AND I have a problem getting my FMS receiver to Enable/Disable the unit, so I'm looking at more tests later in the week around my local corn field..

/Icebear
May 02, 2006, 08:47 PM
Closed Account
Thread OP

El Mirage Dry Lake Bed Test 1


UAVers,

The first autopilot test got off to a slow start when my wife surprised me with news of my daughter's eye doctor appointment (I had to take her). So much for dawn patrol! My friend Nick and I didn't get to the lakebed until noon (read: hot and windy).

The "neat" news was that "someone" was testing what looked like a Predator RQ-1L (might be smaller, might have fixed gear). We could watch the entire flight path including the runway flybys at 5 feet off the deck. Two chase planes (one a single engine Beechcraft of some sort) followed with one in close pursuit and one high up for bird's eye observation. They followed the UAV in probably a ten-mile oval pattern for HOURS. The uncool part was the altitude; well below the FAA's 500' (El Mirage is a State of California Public Recreational Area). The UAV team was not flying out of the Aldelanto Airstrip but a private facility about two miles NW. The buildings seemed too "commercial" to be a part of (near by) Edwards Air Force Base so I guess it is a private test strip. None-the-less, skimming a public playa with a military UAV (a killer BTW) isn't cool. Fortunately, they seemed to alter their flight plan to a much higher northern route when they saw us setting up. I'm sure they have some sort of FAA waiver but how is that going to help when they suck up some kids foamie?

My test seemed anticlimactic after all that! I programmed the Waypoint route into the plane and my handheld (Garmin eTrex Vista). We positioned the plane at altitude and engaged the autopilot. All seemed to work except the ALTE gain (50%) is too high as the plane pitched up and down (more so during turns). The wind made this a tough test to gauge the flight characteristics so we made a hotel reservation in Aldelanto for Wednesday so we can REALLY do dawn patrol on Thursday morning.

Anybody have any PICOPILOT setting suggestions other than that which can be found on their FAQ?

Jeff Parisse
www.teslacoil.com
May 02, 2006, 10:13 PM
Professor of Wood
kd7ost's Avatar
I use the Alt3 E myself. I have the rate gain set pretty low. I have a slight amount of upthrust in my engine so keeping full throttle on won't keep me at altitude with my current gain setting. It will still rise slowly. I have GPS and video overlay system so I can watch the altitude. I have my whole system set up so that at crusing speed, (throttle setting) the Pico Alt will keep me where I set it at. For personal preference, I have my platform set to climb slightly under full power.

I would note those issues in your own plane. If you set the throttle at 50 percent or so, trim it to fly straight and level, then advance to full throttle, does the plane go up or come down or stay level? If all of that is working for you to stay level, then enable the unit and watch what happens. If the plane notably porpoises, you have too much gain. By the same token, if it won't hold it at altitude when going down wind, (sinks) or going upwind (rises), you need a little more gain. It's a balance. Easy to get when winds are calm. High winds will allow the plane to rise and sink in a bigger altitude window. It will likely take a few sessions in different winds for you to dial it in the best for you.

Dan
May 02, 2006, 11:46 PM
Closed Account
Thread OP
Dan,

Thanks for the input. Getting a starting point or other such advice from the PICOPILOT manufacturer always resulted in the same advice: read the FAQ. I'm so glad to be on this fourm with you folks and gettin' some work done (before I have to go back to work).

Given today's prevailing winds, a Waypoint survey by car and your comments, I'm going to set-up an up-wind zig-zag pattern for Thursday's test. That way the wind direction will be simetricall for both left and right turns. Today's Figure 8 pattern was programmed offsite (in LA) and was "askew" to the wind when we got there (El Mirage Dry Lake).

Jeff
May 03, 2006, 11:01 PM
Registered User
Good luck tomorrow morning. It looks like your testing right around where I use to live(Ridgecrest).
May 04, 2006, 03:05 PM
Closed Account
Thread OP

El Mirage Test 2


UAVers,

First the Predator. It WAS flying from Adelanto Airstrip and probably was the Minuteman Project practicing landing approaches. Luckily they werenít flying today (at 7:00am).

Nick and I got to Liberty Point on the Playa at 7:00am and setup. We were in the air a few minutes later ready to test waypoint sequencing. I wanted to observe a turn close up so I made the SECOND waypoint above our position. I, however, programmed the first waypoint almost a mile away. The lakebed looked so much smaller on the computer screen. Even in real life, the illusion of distance is pretty strong. Bottom line: As the plane flew away from us and towards the first waypoint, it seemed like it was NEVER gonna turn. Even though Nick was watching the plane with binoculars, the distance was too far for me to see well so I aborted the autopilot thinking it had over flown the spot and was headed for Las Vegas.

The two flights did give us the chance to dial down the pots and get better turns but we ran out of time and juice before we could load up another set of points. Instead we erased all and tested the auto return feature. It worked great and the pitch problem is resolved. We logged about 22 minutes of airtime.

What we learned: When testing, make the FIRST waypoint close to the observer so one can observer "autopilot on" when the plane veers towards the observer/pilot. Then one can observe "autopilot sequencing on" as the plane moves on to the second waypoint. I'll make my route legs much shorter too (maybe 1/2 mile each?).

Dumb Mistake: I forgot to turn off the rudder gyro for the last two days of testing. The rudder might have been fighting the autopilot during turns. I'll set gain to zero for the next round of flight-testing.

Jeff
www.teslacoil.com
May 05, 2006, 03:02 AM
Registered User
icebear's Avatar
Thanks for the update Jeff! Interesting to read about your testing. I totally agree about setting up the waypoints closer when testing. I had the same issue when testing my first UAV last year. The waypoints where so far away that I "chickened out" fearing that I would loose radio range when my 42" Super Impress turned into a small speck just 1/2 mile away. My longes routes are only about 1.5 miles but the plane is never more than 1/2 mil away.
With my new bigger airfame (SuperMiss, 54")
I'll be more daring. Hopefully I'll be able to test it with the Geko 201 and PDC-10 this weekend!

/Icebear
May 05, 2006, 12:17 PM
Closed Account
Thread OP

El Mirage Photos


UAVers,

Here is a link to the Google Sat. Photo of El Mirage Dry lake bed.

http://maps.google.com/?ll=34.637728...09692&t=k&om=1

Nick took photos of yesterday's test and I'll add them to this post later today...

Jeff
Last edited by workshop; May 06, 2006 at 01:54 AM.
May 05, 2006, 11:30 PM
Registered User
lvspark's Avatar
I look forward to seeing them .. Cool Project!
May 06, 2006, 02:25 AM
Closed Account
Thread OP

Waking the Perimeter


Good luck this weekend Bjorn... It'd be interesting to hear how it goes...
Thanks LVSPARK, it's been fun so far...

San Pedro Flight Tests
Now that the NAV2 and ALTE gains are adjusted for smooth flight and reasonable turns, I'm going to bring the testing closer to home (three hour trips, three dollar gas and hotel rooms were adding up) and fly at a large soccer complex in San Pedro.

Flight Fifteen
Today I learned that it's probably a good idea to walk the (flight testing) perimeter with a handheld GPS to mark the route and landmarks in Mapsource before proceeding with programming waypoints. It seemed that the waypoints I programmed in the night before didn't correlate with the actual field and I couldnít seem to enter the pattern. Once I walked the whole field complex (making notes of landmarks) and loaded the "bread crumb" route into Mapsource, I could much more clearly see where I needed to fly and where I had made my mistakes. The handheld GSP has become essential tool.

I've set a figure eight pattern with the waypoints at least 500 feet from each other. The turns are all equal to or greater than 90 degrees. I hope the PicoPilot has enough resolution to make this route or I'll have to go back to the desert.

Jeff
May 06, 2006, 09:33 PM
Professor of Wood
kd7ost's Avatar
Sounds pretty cool Jeff. You're making some great headway. If your plane can't make the turn it might be more of a factor of aircraft speed and turn rate ability. Your rudder gain needs to be dialed down so the plane won't zig zag and it sounds like you've done that well. If you ask it to turn too sharply it's like trying to turn a 90 degree turn in a boat. It will side skid (slip) as it comes around. If your next point is inside the crafts ability to turn you will overshoot. But wait, there's more.

The Pico Pilot Nav unit uses waypoint departure sequencing which is much better in a small UAV than waypoint arrival sequencing. The "way point made" in the pico pilot sequencing software happens once the unit is within .1 mile of a waypoint. .1 mile is equal to 528 feet. In other words, If you approach a waypoint, once you are within 528 feet, the waypoint will be made good as far as the sequencer. But, it now looks for departure. As the plane travels past the waypoint the distance begins to increase. That's what the sequencer sees to tell it to now start navigating to the next waypoint. You might be 500 feet off to the right or the left of that waypoint if you have a cross wind for example. It will look like your plane didn't make the waypoint to the gnats hiney. It didn't. But it made the waypoint as far as the software goes. This needs to be like that. The small amount of error in a GPS position fix, coupled with aircraft flight dynamics and winds might have you flying around a waypoint till fuel runs out if the spec were too tight. If you had to get within a foot, the plane might skid around up there forever and never make the waypoint. Parameters in the code have to be big enough to allow small light planes to fly a course. The smaller the course gets, coupled with the bigger and faster the plane becomes, the more you can expect to see what might look like erratic behavior when flying a small course.

The plane makes one waypoint and starts to turn to the next. The first one might look pretty good. The plane starts its turn and winds or over correction can cause the plane to turn too little or too much. It starts to correct the other way. These are subtle and look fine from your ground position. Just a little way into the turn though, itís already made the next waypoint as far as range because itís only 500 feet away and the sequencer made it at .1 mile. You might not go right to the waypoint. It might skirt the edge of that round range target and begin to gain greater distance. This tells the sequencer to go to the next waypoint. It cranks over to cross your figure eight and your thinking, man it didnít even come close to that second waypoint. You might think it missed but it made it and moved on already. This isnít always going to happen the same way in such a short and tight turning course. How you enter the course in autonomous mode (bearing and range) can cause the turns to be a little different each time.

Here's an old thread where Icebear and I hammered out a learning curve on the pico pilot. Much of the information would be important to you in your route sizing application. Anyhow, itís just something to be aware of. Icebears original unit used arrival sequencing which made his look really like it was failing when it wasnít. The plane in the early version would start the turn as soon as it got within .1 mile. He had a course smaller than that range so it was sequencing through the course without really getting away from him. The current Pico Pilot variants use the departure sequencing like I described so it is better. If it doesnít look good to you during flight, it might be better to run an oval course with a few in between waypoints to allow the turns to be gentler. That will allow you to evaluate the operation without over steering the plane in a tight course. A slow little plane will stay tighter than a bigger fast plane. Keeping your aircraft speed down to just enough to keep moving along will also help it not overshoot.

https://www.rcgroups.com/forums/showthread.php?t=410349

Dan

Edited to include,

Post 29 and 30 in that thread talk about the distance to waypoint issue.
Last edited by kd7ost; May 06, 2006 at 09:46 PM.
May 08, 2006, 02:24 AM
Registered User
icebear's Avatar

Second testflight


Interesting read Jeff and Dan - and I am looking forward to the next flight report.

In the meantime I managed only one testflight of the Supermiss this weekend due to strong winds. I am planning to test the PDC-10 + Geko 201, first "return home" and the a short route.

Unfortunately, when the winds calmed down sunday evening and I finally could walk to my nearby field (I am lucky to have one really close), the FMA co-pilot wasn't correctly setup. The plane behaved well but when engaging the Co-pilot I had set the gains to low and it would loose height and enter a shallow spin. Now everything is re-calibrated and ready to go this evening. We have 75 degrees F and calm winds!

Here's a small picture of the route.

/Icebear
May 08, 2006, 10:54 AM
Closed Account
Thread OP

Thanks Dan!


Dan,

Thank you so much for taking the time to describe waypoint sequencing to me. I have a much clearer picture in my head on how PICOPILOT works and now I can see the how the map and plane relate.

MAPSOURCE was planting a visual image of a straight-line path. That's the way the route looks on the screen. Given your post and the dialogue with Icebear, I can see now that instead of a line, I should think of a 500' wide swath (representing the margin of error) with the plane overshooting the WP before "looking" for the next one. I should also slow down.

Here's a before and after of the next test I have planned for today...

Thanks again...

Jeff
May 08, 2006, 06:10 PM
From Sky Colombia
libelulamodelos's Avatar
i got also a telemaster , now our team are developing a "TERRACO", pictures soon at:
http://andresherreracali.blogspot.com


Quick Reply
Message:

Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Sr. Telemaster/add ons mike mcfadden Fuel Plane Talk 23 Mar 24, 2010 08:00 AM
Sr. Telemaster Alan H. Siegel Aircraft - Electric - Airplanes (FS/W) 0 Jul 07, 2002 09:46 AM