anyone using Crossbow uNav?

[Unterhausen]

I'm new here, although I've been lurking and reading for quite a while.

I'm at Penn State, and we have delt with Crossbow for their sensor node computers. We have been wanting to get into UAV's for a while, and bought model airplanes, but have been holding off on getting an autopilot. We have looked at the Piccolo and the Micropilot, but I can't see spending that much money (must admit I'm pretty close to buying a micropilot).

So when I saw that crossbow had an autopilot for $1500, I bought it. It has GPS and all the other flight sensors. Their manual is less than helpful. The nice thing is the source code is included. I was wondering if anyone had any links to people using the uNav?

The thing we are struggling with right now is the RC receiver hookup. It wants a ppm input, and we just have PCM receivers. I'm trying to figure out which receiver to buy, and once I buy it, how to hack it so I get the ppm signal. Is it possible to buy a receiver that outputs PPM without hacking into it? Paparazzi has some images of a hacked receiver, but I didn't find anything more.


I'll try to post our experiences as we progress.

[vpatron]

Hi, neat that you get to play with this stuff in school!

I've been thinking about this since I'll need to pull raw PPM data to process into my own controller at some point.

I would take a simple but good RX like Hitec micro 555 and look for some sort of FM demod chip, and trace the circuit after that with a scope and look for a raw PPM signal. I was thnking that the Hitec 555 would be good because it's a very reliable full-range dual-conv RX, plus its dumb: there is no micro doing fancy processing. You can buy used 555 for $35 here on RCGroups or Ebay.

I would guess there will be some opamp or comparator to square up the demod signal then to a multiplexer to split out the PPM channels.

Of course you need to have a PPM TX on.

Let us know how it goes because I need to do this at some point, too.

Regards,

-Vince

[Myron]

Got a link to the Crossbow?

Myron

[kd7ost]

Perhaps we should discuss this in greater detail and pull in some of the guys who have worked with it. As I understand it, The output from a receiver to a servo is a PPM signal irrespective of whether it comes from a AM, FM, or PCM, (which is still FM) Radio. The detriment to some UAV guidance systems in a PCM radio is the processors "failsafe mode". Some autopilot systems want to see things go dead from a standard receiver where as a PCM receiver, if enabled, will put out a preditermined PPM signal or continue to send the last known PPM signal during "Failsafe" operation if the transmitter is shut off.

I don't know the Crossbow or Micro-pilot systems Here are some more good choices just to muddle your process. ;-)

http://www.uavflight.com/
http://www.cloudcaptech.com/

The UAVflight system wants to have a dead receiver when enabled according to an e-mail I got from them asking about that very issue. They recommend a standard FM transmitter and receiver to get rid of the Pulses that a PCM system will still continue to generate.

Some PCM systems allow you to turn off the PCM and use it with a standard receiver which will go dead as needed. In that case I think you have to purchase a suitable matching receiver for your transmitter that does not have the failsafe processing functions built in. I'm not sure if thats the case will all brands and/or models.

My whole issue is probably moot, I think your asking whether off the shelf autopilot systems use failsafe, or go dead, when the transmitter signal is removed?

Dan

[Unterhausen]

Got a link to the Crossbow?

Myron
Sorry, I was going to post a link to the uNav.

uNav (http://www.xbow.com/Products/productsdetails.aspx?sid=133)

This is not an autopilot. It is a AVR that provides an interface to the magnetometors, accelerometors, gyros, pressure sensors, and gps. So to actually implement a flight control, you need another computer. This done over serial connections. If you use the Stargate (http://www.xbow.com/Products/productsdetails.aspx?sid=85) , it connects directly, but it still is using the serial ports. The Stargate is a XScale board sold by Crossbow, but designed by Intel. It got some press a while back since Intel was sponsoring universities to use it for robotics research.

I learned a bit more about the hardware since I posted. Part of the problem is i haven't done much with RC planes. It turns out that the uNav monitors channel 4 of the RC receiver and if that goes over a certain length, it knows to stop autonomous control. I guess I could probably figure out where the ppm signal is on a receiver with an oscilliscope. The PCM receiver we have has a separate RF board, and then there is a chip that looks like a microprocessor that demultiplexes and translates to the pwm signal for the servos.

I saw a note on the Paparazzi site that said they wanted to reverse engineer the PCM signal, and use that for manual control, so I suppose I could do that as well. Didn't check to see if he had any code yet.

[Unterhausen]

My whole issue is probably moot, I think your asking whether off the shelf autopilot systems use failsafe, or go dead, when the transmitter signal is removed?
Dan

Well, I wasn't that far along til I read your post :)

The Crossbow system is not really a failsafe, it's a way of swapping between manual and autonomous control. I'd have to think more about failsafe operation. It seems to be a fairly serious concern. From bench testing, it appears that in the absence of a signal from a transmitter, the servos go to a default position. They are using the output of channel 4 to select between manual and autonomous control. For my applications, the absence of a signal from the ground should trigger a failsafe condition.

A PPM receiver has the PPM for each channel time multiplexed. There is a blanking period, then one pulse each for Channel 1 through 8, and repeat. So the receiver just demultiplexes the signal with no modification. This is the signal that Crossbow wants. PPM also sends a pulse width associated with zero. Negative pulse widths are just shorter than a zero pulse. I like the Crossbow approach, and actually there is no standard for PCM, so they have adopted the most standard approach. We just bought the wrong kind of receiver. Our transmitter works with either PCM or PPM receivers. And there apparently is no receiver you can buy with the PPM output. PPM is also used to make the trainer connection between transmitters. I thought I might see a receiver with that output, but no joy.

Micropilot (http://www.micropilot.com/) has a pretty nice system. Like Cloudcap, they have ultrasonic automated landing, which I like. They have a pretty good deal for educational users. The thing about the Cloudcap Piccolo is that it costs at least twice what the Micropilot does and goes up from there. And the case that it's in makes it weigh too much.

[kd7ost]

Our transmitter works with either PCM or PPM receivers. And there apparently is no receiver you can buy with the PPM output. PPM is also used to make the trainer connection between transmitters.

OK, So using the term PPM you're describing the serial pulse train from the sync pulse through to pulse 8 (or however many your system transmits) And you want to get at that pulse train prior to it being demuxed back out to the individual channel or servo ports?

Most receivers, be they PCM or PPM have that train just before a final output stage in a receiver. I know you want to get rid of the PCM failsafe function as that would interfere with your autonomous systems. You can take a standard receiver and find where the pulses come into that final output demux, devider, ripple counter whatever it is. Commonly it's a D type flip flop ganged together to route the pulses to the correct output. I'm not sure if I'm throwing out what you're looking for bvut you should be able to get to that full length stream right near the servo outputs.

Dan

[vpatron]

Dan, yes, I think that's what he's looking for: the raw PPM with blanking pulse and all 8 channels on one wire, before the demux. 73 de kg6izl... :)

Unterhausen, I really don't think it's too hard to find if you take a Hitec 555 look for a demux chip and start poking around with a scope.

Another idea is to just ask Peter Berg and get one of his receivers. He has sold the design and production of Berg receivers to Castle Creations last year and CC is not yet in production, but I'm sure you can beg for a used one in the classifieds section of RCGroups if Peter will give you the info. (Maybe he'll sell or give you one?) His user ID is "potifar". He's a pretty nice guy.

Regards,

-Vince

[kd7ost]

Heya,

73's vpatron.

Unterhausen,

That final chip used to demux should have a reset line on an input too. You can likely read the pn off the chip so can just look up which pin is the serial input. Of course a scope will prove it out to those of us that just have to see it anyway.

Dan

[sesat]

If you haven't found the PPM train off a Hitec555 or ELectron6, pLease continue posting here.

Ram.

[Unterhausen]

Thanks for all the replies. Are the Hitec or Electron6 going to be compatible with a Futaba transmitter?

[vpatron]

Unterhausen, as long as your Futaba transmitter transmits PPM (not PCM, read your manual) and is on 72 MHz band, it will be compatible *IF* you have the right "shift" receiver.

Other newer RX are auto-shift: they listen for the TX on power up and configure themselves appropriately. But for the 555 and Electron, you have to make sure you order the correct shift: either positive or negative shift.

Futaba is negative shift which is same as Hitec transmitters. JR and Airtronics transmitters are positive shift.

See http://www.hitecrcd.com/Receivers/electron6.htm

-Vince

[Bg~]

I'm at Penn State as well, working on a group of UAV's using the Piccolo as it happens (with ARL). Our plan is multiple UAV's flying cooperatively. I saw the uNav stuff at AUVSI in June, and was impressed. I don't have any practical experience with it though. On the micropilot, I have the opinions of 3 of the AUVSI student design teams that it is a pain to get working properly (one of which is a team from my undergrad school). On the other hand, a guy from NASA gave a presentation in which he completed something like 250 flights on the micropilot, and 30 damage free autolandings, so maybe its just the students (I'm a grad student also).

BTW, what department are you in? Maybe we can compare equipment sometime.

PS. Cloudcap may offer you a discount if this is a student project. Micropilot may also do this as well. Can't hurt to ask, they've done it in the past for the student competitions.

PPS. About the automatic landings: From what I gather, they are dodgy. You need a study airplane because you will take some hard landings. The guy from NASA's ratio of damaged to damage free autolandings was pretty high with the micropilot.

[AnthonyRC]

Something tells me that a picture here is worth a thousand posts.

I use the Hitec electron 6, and modify it with a very simple change to route the ppm through to channel 6. Works a treat!.

Thanks to Rolf Z for finding this mod!

[Unterhausen]

I'm at Penn State as well, working on a group of UAV's using the Piccolo as it happens (with ARL). Our plan is multiple UAV's flying cooperatively.

I'm also at ARL. I don't have much hope of flying multiple planes in cooperation, but we'd like to have coordination with our robots. I think we may have the same planes, because Al Niessner came and talked to us about how to get started. My boss shot down the Piccolos because of how much they cost, but maybe we should get them for commonality. My email is eek105@you know where. One of the problems is that none of us fly, so recovery seems like it might be problematic.


AnthonyRC, thanks for the pictures. I'm going to go see if the local hobby shop has that model, but now the process of reverse engineering a different reciever is pretty obvious to me.

[Unterhausen]

I finally got motion out of the uNav. I tapped into the Futaba transmitter trainer port, switched the trainer port to work with PPM, and fed that directly to the uNav. Turns out that Crossbow the windows gui test program that comes with the uNav has the servo ports randomly numbered, not the same as on the hardware.

So the behavior that we have observed so far is that the uNav will not move the servos unless a PPM signal is coming in on CH4(?). CH4(?) determines the servos are under autonomous or RC control. If there is no PPM signal, the servos go to a default location. It sounds like a fairly reasonable setup, although as far as I can tell the default position is not configurable without a recompile.

[vpatron]

Anthony, good info. Thanks!

Has anyone done a Hitec 555? I can do it when I get there, but thought someone might have saved me the trouble. :)

-Vince

Something tells me that a picture here is worth a thousand posts.

I use the Hitec electron 6, and modify it with a very simple change to route the ppm through to channel 6. Works a treat!.

Thanks to Rolf Z for finding this mod!

[Unterhausen]

I had found a reference to Stanford flying the micronav, but I looked at this link http://airtraffic1.stanford.edu/~uav/index.html and they are using one of the more expensive Crossbow INU's

[Sauli Sarkka]

Anything new been done with the uNAV + Stargate? We are probably getting a set for a quad rotor project.


-Sale

[clolson]

xbow just released v1.4 of their kahlman filtering/autopilot code. I had done a lot of clean up based on v1.3 but they made substantial changes to their kahlman filtering ahrs/nav code which means I either do a *really* tricky back port to my code (probably with lots of errors I'd never be able to debug) or start from scratch based on v1.4. I'm hoping I can talk them into some small restructuring of their code to facilitate the job of people like me who wish to develop much more flexible and advanced flight controllers.

Curt.

[Sauli Sarkka]

Any more specific information regarding the changes in the new Kahlman code compared to the old one anywhere?


-Sale