I’m considering adding a GPS to my BlueROV2 like Kevin did. Anyone have any suggestions for a low cost, small form factor, centimeter accurate GPS? Found this, but it’s not in production yet: http://diydrones.com/profiles/blogs/tiny-rtk-the-world-smallest-precision-gps-and-the-cheapest#comments
I got a close up look at a commercial UUV last week and that had a very tiny Taoglas patch antenna molded into the dorsal fin. I’d probably estimate 10x10mm or less. I’m not sure on the exact model, but I’ll keep looking through the resources and let you know what I find.
From the PX4 project, here is the list of compatible GPS’: https://pixhawk.org/peripherals/sensors/gps
From that, I had been thinking about something like this: http://www.csgshop.com/product.php?id_product=170
Centimeter accuracy GPS units are not very common yet and they are expensive. There are tons of small and cheap GPS units that can do meter accuracy. Is there a reason you want centimeter accuracy for the ROV?
Realistically speaking, 1 meter accuracy is probably good enough for now (and the short term), but how often are they really accurate to 1 meter? Don’t they all depend on the quality of satellite signals? Is it realistic to expect 1 meter accuracy from a GPS with an antenna only inches above the water’s surface?
From my experience they’re usually accurate to 2-3 meters. Accuracy depends on the signals but also on the processing method and the application. I haven’t tested myself so I’m not sure how much the water surface affects it! We’d have to ask Kevin K or someone else who’s tested it!
On it tomorrow night, now that I have a compatible computer! What metric would you like me to test with the stock 3DR GPS?
Does ArduSub (or ArduPilot for that matter) offer a “hover” mode? If a quad-copter’s GPS is only accurate to 2-3 meters, how well does it maintain position? Does it allow a certain amount of drift before it issues a correcting movement?
Hover mode is generally within a few feet or less on multicopters. They do however use the accelerometers and gyros in conjunction with GPS and compass to feed into the position and hold logic.
Most newer GPS units for Pixhawk support both North American and Glonass which offer better coverage and satellite lock as well that helps with accuracy especially places where low on horizon may be hard to get enough satellites for good 3D lock.
Hi all, I am tech support for survey grade GNSS equipment and have learnt more about GNSS than I ever wanted to . There is a reason high accuracy GNSS costs so much. Basically any L1 only GNSS unit (even if GPS/GLONASS/Beidou capable) is only going to get you to +/- 3-5m stand alone. To achieve sub metre, you need RTK. This involves a base station set up and the correction data is transmitted to the rover to apply the error modeling to give you a sub metre answer relative to your base station. There are SBAS (Satellite Based Augmented Systems) out there that can send corrections to give you sub metre or even decimetre but they are subscription based, and not cheap. There are lots error sources that can affect the solution. Multipath and Ionospheric activity are 2 of the biggest factors which can be reduced by using the L2/L2C/L5 etc signals and good antennas with built in ground planes in them. Something none of the cheap units have. I have trialed cheap units against $60k units and sometimes the answers are close. The expensive units give you repeatability and also good heights. Something the L1 only units were unable to achieve. There are cheap RTK units like the Reach RTK (emlid.com) for example which will give you quite good results, definitely in the sub metre. Aim for sub metre with your GNSS and if you get better then great, otherwise you will get what is realistically expected.