Free World (or at least European :-) ) Map Data for 3D visualisation?

[FirmamentFX]

Hi guys,

What are the options out there for free world map data to aid in a 3D visualisation of a UAV mission?

I am thinking about using a 3D world map for visualisation beyond video range for my UAV (ie when I will be using data fed back from the UAV only rather than a video feed).

What is the largest scale free data available (without needing hundreds of hours of processing)?

I was looking at the FlightGear website. They make the source available, so potentially I could use that for the system. I was wondering what other people thought?

Many thanks,

Martin

[clolson]

I was looking at the FlightGear website. They make the source available, so potentially I could use that for the system. I was wondering what other people thought?


I think FlightGear is a great idea, but I'm somewhat biased being heavily involved in that project. :-)

FlightGear has been used to visualize UAV flights both in real time and later replaying recorded flight data and it works quite well. FlightGear allows you to view your flight from a variety of vantage points. If you feed in enough data, you can start making virtual instrument panels come to life, virtual HUD, animate control surface deflections, add audio effects, etc. There is a lot of fun things you can do with a little bit of effort.

FlightGear has a rich set of network interface modes so by writing a little bit of "glue" code that can take your UAV data and reformat it for FlightGear, you can visualize your flight with a "stock" copy of FlightGear and wouldn't need to touch any of it's code.

I've posted this link before, but scroll down to the December 12, 2005 entry at the following site to see a movie of a FlightGear view side by side with the real camera view.

http://baron.flightgear.org/~curt/UAV/Rascal110_2/

Regards,

Curt.

[FirmamentFX]

Hi Curt,

Thanks for the reply. That video looks very interesting - especially the blended overlay!

My plan was to create a system whereby a "real" video feed, a 3D map, and a HUD could be blended/mixed at will and have a second monitor giving an instrumentation view (PFD, MMD - possibly - and other instruments and multi function displays).

Can all these be programmed into FlightGear, given the open source nature of it?

Obviously you are biased ;) , but do you find the 90m accuracy to be enough?

How about mixing it with GeoVML models of built up areas? Has that been tried at all?

Cheers,

Martin

[FirmamentFX]

Sorry - what about weather data? It would be cool to be able to overlay a kind of "faked" weather radar from data downloaded in real time from the net?

[clolson]

Thanks for the reply. That video looks very interesting - especially the blended overlay!

My plan was to create a system whereby a "real" video feed, a 3D map, and a HUD could be blended/mixed at will and have a second monitor giving an instrumentation view (PFD, MMD - possibly - and other instruments and multi function displays).

Can all these be programmed into FlightGear, given the open source nature of it?

Obviously you are biased ;) , but do you find the 90m accuracy to be enough?

How about mixing it with GeoVML models of built up areas? Has that been tried at all?


Superimposing the camera over the synthetic view is an interesting way of quickly/visually determining of your IMU data is in the right ball park. These IMU systems are typically plagued by a lack of "truth reference". They system says you have n degrees bank, m degrees roll, heading of xxx, but how do you validate that? Comparing to video at least lets you know if you are close. It's fun to see all the little wobbles in the video match up with corresponding wobbles in the IMU data.

For blending all the data into one application ... I think that can be done. You will need to do some work with frame grabbing and converting each video frame to an opengl texture so it can be rendered inside flightgear ... but that is all doable ... not a lot of magic there, just some leg work.

90m terrain accuracy is pretty good from an aerial perspective ... at least for full scale aviation. If you are operating a uav in tight quarters, or flying a helicopter, maybe you need more detail and more accuracy. But if you can find more accurate data, please let me know. I'm not aware of anything better than SRTM that has any kind of wide area coverage.

I have no experience with GeoVML, but you can drop a variety of differently formatted models into the flightgear scene pretty easily. That is how we did our flying area ... took a screen grab from a certain mapping site, dropped it on a flat square, added the hangars, saved that as a single 3d model and placed it in the scene a meter or two above the default terrain for the area. It actually works quite well and gives us a "photorealistic" model of our flying area.

Curt.

[FirmamentFX]

90m terrain accuracy is pretty good from an aerial perspective ... at least for full scale aviation. If you are operating a uav in tight quarters, or flying a helicopter, maybe you need more detail and more accuracy. But if you can find more accurate data, please let me know. I'm not aware of anything better than SRTM that has any kind of wide area coverage.

I am not aware at the moment of anything better...

Using a hybrid GPS/INS/radio altimeter system would help with tight quarters. I would like to look at using a commercially available mini radio altimeter looking forward and down (at an angle) as a "rough and ready" terrain following tool.

If you know the angle of the RA, and your altitude (from a downward looking RA) and the AoA of the aircraft at ay given moment (from an INS system) you could trigonometrically (sp? in fact, it that even a word?!) work out what you need to do to avoid the next obstacle. Then it just comes down to response time...

Of course you probably wouldn't want to go down to 20ft with this system, but it could be interesting.

In conjunction with a "knowledge" of the terrain using SRTM data, it could be fun! :)

Will reply to the rest of your post later. Have to dash out for a bit!

M

[clolson]

I've posted this link before, but scroll down to the December 12, 2005 entry at the following site to see a movie of a FlightGear view side by side with the real camera view.


It just occured to me that I could just as easily upload these two videos to youtube and it would probably be easier for everyone to see them ...

Side by side:

http://youtube.com/watch?v=SAMGnK9ztdA

Blended:

http://youtube.com/watch?v=RgK1019Bjno

Regards,

Curt.

[Unterhausen]

Curt, that's really impressive. Looks like you may have your biases set a little off. Is that using your KF code?
Eric

[clolson]

Curt, that's really impressive. Looks like you may have your biases set a little off. Is that using your KF code?
Eric

Hi Eric,

This was done with a Microbotics MIDG sensor on semi-permanent loan to us. It's pretty much as nice as you can get for $5k with really well developed KF algorithms. You can notice places where the heading estimates drifts from reality, but based on what I'm learning about these things, getting a good heading estimate is *really* hard to do.

I don't have anything like this to show for the MNAV + my own code just yet, but that is in the pipeline. I'm collecting bits and pieces and making cables right now to mount the MNAV + a gumstix in an existing electric powered senior telemaster. I'm thinking that maybe the week after next I might be ready to fly that with the MNAV/gumstix in the critical control path.

http://baron.flightgear.org/~curt/UAV/SeniorTelemaster/Large/IMG_4445.JPG

I know I keep saying this all is going to be cooler than sliced bread when I get it working, but dangnabbit ... (sorry, the wife has been making me listen to too much country music in the car) ... doing this stuff in your spare time, with a family, house, dogs, full time job, etc. ... things just don't happen as fast as they should. :-(

On the plus side, there's a good chance I'll have taught myself how to solder by the time I get something flying autonomously. :-)

Curt.

[reedchristiansen]

Curt,

Flightgear sounds very nice. I would like to replace our current in house simulator for hardware in the loop with our autopilot with flightgear. By what I have read on this forum and other places it lookes ideal.

However, I have spent several hours digging and cannot find any documentation on how your interface with flightgear over TCP. I assume it is either UDP or TCP/IP with control surface deflections and other things going to flightgear and state information going out of flightgear.

If you could point me to some documention or example code, that would be great!

Thanks,
Reed

I think FlightGear is a great idea, but I'm somewhat biased being heavily involved in that project. :-)

FlightGear has been used to visualize UAV flights both in real time and later replaying recorded flight data and it works quite well. FlightGear allows you to view your flight from a variety of vantage points. If you feed in enough data, you can start making virtual instrument panels come to life, virtual HUD, animate control surface deflections, add audio effects, etc. There is a lot of fun things you can do with a little bit of effort.

FlightGear has a rich set of network interface modes so by writing a little bit of "glue" code that can take your UAV data and reformat it for FlightGear, you can visualize your flight with a "stock" copy of FlightGear and wouldn't need to touch any of it's code.

I've posted this link before, but scroll down to the December 12, 2005 entry at the following site to see a movie of a FlightGear view side by side with the real camera view.

http://baron.flightgear.org/~curt/UAV/Rascal110_2/

Regards,

Curt.

[FirmamentFX]

Reed,

You took the words right out of my mouth (or, the strokes right off my fingers technically :) ). That was my next question!

From my point of view, it might be more suitable for me to create a "custom" version of FlightGear rather than "hacking in" to a pre compiled version. I am not going to need bitmap texturing or detailed aircraft dynamics handling for example. The idea is simply to visualise data that is being sent from the real UAV as a 3D moving map and a series of instruments.

I assume as FG is open platform it is not dependent on the .NET framework / managed code in Windows? If I do my own compilation, I may well compile it for CE and run the ground station from CE as well as the UAV - at $16 per unit ($3 for the standard version) CE is a lovely hard real time OS packed with features for the price...

Cheers,

Martin

[clolson]

Real quickly, FlightGear has a couple options --native-fdm=param1,param2,... and --native-ctrls=param1,param2,etc.

The fdm structure is all the flight and location parameters. The ctrls structure is the control inputs.

Using those you can setup inbound or outbound socket connections, either udp or tcp. You can set the data rate, port, host, etc.

In the code we provide "public domain" definitions of the corresponding structures so you can quickly add support in your own code to send or receive these structures.

Curt.

[reedchristiansen]

Do you have any example source which populates the structures and then sends the commands to flight gear, and which also retrieves populated structures from flight gear.

Thanks,
Reed

Real quickly, FlightGear has a couple options --native-fdm=param1,param2,... and --native-ctrls=param1,param2,etc.

The fdm structure is all the flight and location parameters. The ctrls structure is the control inputs.

Using those you can setup inbound or outbound socket connections, either udp or tcp. You can set the data rate, port, host, etc.

In the code we provide "public domain" definitions of the corresponding structures so you can quickly add support in your own code to send or receive these structures.

Curt.