alternative to GPS navigation

[ryan_beall]

I would like to increase the capability of my homegrown uav system with out the use of gps. I am using an IMU as well as pitot statics. I validated the code as the hardware last summer using gps. My goal is to radio direction find using doppler shifts from radio/tv signals or cellphone towers or anything really. I am up for any lightweight (1lb) robust method of non-gps navigation. I have read some and seen what the kestrel does with video recognition navigation but I would like the system to be as "pilot out of the loop" as possible while opperating with out gps.

Any robust and proven ideas would be great
Thanks Ryan

[zlite]

The two best methods I know of are "visual flow" (for aircraft) and lidar (for helis). But the doppler shift sounds interesting, although I wonder if it will be precise enough to be useful at low speeds.

[UPDATE: my apologies; I didn't read your question closely enough. Visual flow is only used as a replacement for an IMU and Lidar is only useful for ground obstacle/target navigation. Neither will substitute well for GPS.]

[hg1]

It's an interesting question.

A traditional method was to use radio beacons such as Loran-C for aircraft and ship navigation - see
http://en.wikipedia.org/wiki/Non-directional_beacon
http://www.navcen.uscg.gov/loran/Default.htm
http://en.wikipedia.org/wiki/LORAN

As far as I can tell, the Loran-C transmitters are still in operation, and there may be some existing low-cost receiver technology. Actually, the software to decode the signals is probably not complicated.

As regards the use of visual navigation, it's certainly possible, though somewhat more complicated. You basically need a pretty good on-board database of visual landmarks and some complex pattern matching software.

At night, there's always the stars, though position resolution will be in miles rather than meters.

[ryan_beall]

I have a DDF1 from Ramsey that I am going to modify within the next week or so to get some "on the ground" data. But its only good up to like 22.5 degrees (3bit) resoluition. I was looking at also using a mouse type visual chip but lensing problem.

[murmur]

You only want robust and proven? What about cheap, light, and available on any streetcorner (wait, you said 1 pound. What about cheap and available on any streetcorner?).

In the "needs development" corner, there are 3D-image sensors being brought to market by several companies, among them Canesta, Inc.:

http://www.canesta.com/html/development_kits.htm

These could conceivably be used for navigation in proximity to objects, although I don't know what kind of maximum range the Canesta devices are able to support.

Dave

[ryan_beall]

intrestingly enough....I attend the Naval Academy and they have three of them..... I jog out to the occasionally. Pretty sure they don't work any more (loran c that is).

[vector_vortex]

I dunno about doppler and radio signals, you are likely to have issues with diffraction or reflection affecting the accuracy of the measurement, doppler shift measurement will also have issues with any form of modulated signals, FM in particular will cause chaos with your doppler system. Simultaneous Location and Mapping, SLAM is a proven and robust technology used in some cruise missiles. Various people are working on non-GPS scene matching navigation, see:
http://www.military.com/features/0,15240,155328,00.html

Have Fun

Joel

[deh6]

What about an ADF using AM radio stations?

I have thought about this "off-and-on." The last time I went so far as to buy a Coby CX-9 AM/FM radio (about $10), to see if it could be hacked. Without the battery it weighs about 3 oz; take the case off and it weighs "nothing." Just holding it in the hand and rotating it, one gets a very good null (i.e. AM stations since the loopstick has a well defined null). Since it is digitially tuned one could switch between a number of known (both freq and location) radio stations and compute position as well as get heading.

The implementation can range from--
1) simple: mount the radio on a servo and physically rotate it to find the null (there is 180 deg ambiguity). Or, even simpler, in conjunction with a mems gyro for short term heading, just mount the radio so that the null alignment is known (e.g. in-line with the fuse) and turn the plane to find the null. (An occasional 360 could be used to correct drift in the gyro.)
2) more complex: hack the rf circuit and add a loop-stick physically at 90 deg; modulate them at 90 deg with turns that are "shorted" by FETs, driven from the micro. There is an old patent that describes the scheme nicely (if there is interest I could dig out the number). And the ultimate would be to add a electric field ant input to remove the 180 deg ambiguity.

[hg1]

It would probably be easier just to start from scratch with a tunable bandpass filter. Maybe have multiple filters to accommodate multiple stations, and do a 360-degree radar-style sweep with your antenna. The more directional you can make the antenna, the better.

[murmur]

Here's a method (coded for use in Windows Mobile phones, et al) that is claimed to use user-created maps of fixed WiFi transceivers and cellphone towers for positioning via "triangulation". Not entirely sure what kind of data they triangulate on, since there won't be any doppler information (would there?) or any direct time-of-flight measurements available. Their claims for accuracy are not overly fantastic, though (a good thing!):

http://www.navizon.com/faq.asp#accuracy

Dave

[hg1]

Given the stated accuracy (100's of meters), they're probably just using RSSI (signal strength) as a proxy for radius. Still kind of a cool idea ...

[Blue Sky]

Not knowing much about it but thinking there must be a fair number of radio beacons out there, I came accross some interesting things. In the Ham radio community DXing beacons is something of a sport. There are many amature beacons as well as commercial and government beacons. If there ae some in your area and if they have time stamps you could use their exact known`location and timing delays to figure your position, well, quite exactly! With a Ham license you could even build your own beacons and put them in known locations.
Here's one site that looked interesting.
http://www.dxzone.com/cgi-bin/dir/jump2.cgi?ID=5652

-Dave