Excellent video for a newbie getting into RTK's such as myself. Wish I had your tutorial 4 months ago when I started using the C94 M8P. I like how you explain the small details which the guide does not cover, and those are really important for people like me that do not really know whats going on, but really would like to see some results. Looking forward for more videos.
Small correction: In case of GPS, each satellite has a completely different PRN code (pseudo random number). If they transmitted the same PRN then you'd have trouble picking up different ranges (if dopplers were sufficiently close together) due to cross-correlation. GPS uses CDMA so that is the case. This is true though in case of GLONASS which uses FDMA and where all satellites do in fact transmit the same PRN (but on different frequency bands). EDIT: You may also be able to get network RTK corrections (not sure if available where you are). For that you'd need a single RTK enabled rover (no base required) + a cell modem or something to access the internet (even a cell phone). The corrections are broadcast over the internet (generally using NTRIP) and are fed into your receiver which replaces the base. Network RTK in most case will have extended range when compared to single base RTK (sometimes up to 50km from nearest base). EDIT2: FLOAT ambiguity simply refers to the accuracy of the ambiguity resolution. i.e. for satellite X the ambiguity was resolved to be 3.45 cycles which has a floating point resolution and hence the name. Fixed refers to fixed or integer fixed ambiguity and it is more precise; this means that the ambiguities are resolved and fixed to actual whole integers. For most applications (less than decimeter) FLOAT is good enough however you will see a quantum leap when going from float to fixed in terms of stdev. Generally on start up you'll see the rover start with FLOAT and then step to FIXED.
uBlox is an amazing company. I brought a M8U from them (but its only about 70$), but I screwed up the silk screen (on my designed PCB) and mixed up Vcc and GND.... They sent me a new one and wished me good luck with my project :)
There's a nice overview here: www.u-blox.com/sites/default/files/products/documents/GNSS-module-selector_Overview_%28UBX-14000426%29.pdf The M8U includes an IMU
Oh that looks very nice indeed. I guess it would improve the autonomous farm rover application a lot. I'd also be curios to see a multicopter using this and doing a position hold, that should be awesome! Thanks for another very interesting video
At that price point, you would have thought a Polysilicon fuse and transorb would have been in the budget for them to include on the modules! Also, you may wish to locate and purchase some RS422 transceivers for doing serial communications over such long cables. They'll be much more reliable in the presence of interference. Great video!
Yeah, I was worried about that so I compared the input and output signals with my scope. I had to go all the way to 100ns/div to see any difference, so I figure this particular link and cable combination should be safe for a pretty good distance: www.iforce2d.net/tmp/MAP004.png
iforce2d for your setup purposes it’ll probably be just fine, but a car with a noise alternator driving by, a farmer with a walkie talkie or perhaps a nearby electrical storm could corrupt the occasional packet. I guess eventually it’ll be standalone (solar powered?) and do all its work over the onboard telemetry link...
Excellent intel, much appreciated! I'll try to make a base/rover setup for an autonomous lawn mower with obsolete u-blox LEA-5T modules, which can be bought used for less than 20 euros on eBay.
Dumb question. How much accuracy would you expect to get if you had a DIY differential system with (2) M8Ns? (1) base station M8N set on your home's roof or other set location and (1) rover M8N with the base sending offset/correction info via nRF24L01? I would think that the base would send info like direction and distance of the offset.
It's more than just a differential M8N actually. What I think the RTK does is that the base station averages the clock phase in order to extract a very accurate clock data stream of its own. The GPS/GLONASS satellites each have their own Rubidium reference clock from which the data streams are generated. Regular GPS modules derive this clock stream during fix computation from received signals (what is indicated by the "3D"), making it subject to any path variations. I suspect the base station stops doing this once it has established it's own version of the clock and switches to using this (hence the "TIME" indicator). For this to work, the base station needs really good time keeping of its own - maybe having a Rubidium clock, for instance. In the case of the unit being tested, I suspect that's what the blue plastic box is all about - temperature regulation for a stable local clock. Using an M8N as the base station does not give this clock/temperature stability. Assuming the hardware of the M8N and M8P is the same, regulating the temperature of the M8N GPS unit might do the trick. The blue box encloses the GPS unit with it's crystal clock. It excludes the RF transceiver module because that usually sees a lot of temperature variation from the output amplifier.
@Daniel Wee, If your theory is correct... I bet the offset would be as slow moving as the temperature changes, and relativistic anyway. This probably means that for your average 20-minute mission, it won't matter; RTL will still be centimeter-accurate. Ardupilot (and DJI) is mostly all relativistic navigation, cause you're lucky if it lines up with google maps at
I'm surprised to hear that you got a 100m difference in subsequent flights to the same place. You mean when you launched from a different location the second time right? I don't know how other firmwares do it, but in the iNav source code I have seen that it does most of the position calculation in a meters-based local coordinate system with the launch point as the origin. So in theory when you give it a lat/lon target, the location it derives may not be quite the same as when starting from a different origin, but 100m seems pretty outrageous. With a M8N I have rarely seen it more than about 3 meters off the google map image. Even if you doubled that and assumed each origin was off in the worst possible direction, that's still only about 12 meters discrepancy between origins. The math should be using double precision so even at 2 miles away from the origin I would expect a lot better than 100m difference. So if the difference is due to the inner workings of the flight controller math then RTK will not help much. But if it's not, then RTK would work great - just don't move the base station of course.
Great stuff again Chris. The nail and MCU shows why you can never have enough breadboards! I was about to buy a few more from BG but hoped to find a wider (not longer) plastic breadboard with no luck. Anyone know of where to find them? BTW, I think "Hammond" is the name of a large enclosure manufacturer.
hi! great video. I´m a total station surveyor trying start in the gps world. Now it's getting much cheaper and more accurate. Not very convinced by the reach rs solution, maybe i´l dedicate some time to a diy solution. Can you talk a little bit more about your microcontroler solution? did you get to program any datum conversion algorithms , or just simple angle to x,y calculation? btw, love the sound of the birds and other animals on the background!! you are definitelly a fortunate man!
The numbers shown are just taken from the NAV-PVT message, which has lat/lon/hdop etc included. The program is basically the same as what I looked at in detail here, but using a different message type: ua-cam.com/video/TwhCX0c8Xe0/v-deo.html
Hi, I would like to ask a question. I see a very good precision in the final result. Do you use the position calibration required by the Base Station in survey-in mode or fixed-mode? Thank's
Great video! Thanks for the instructions! Is there a way that I can use an Arduino board to read lattitude and longtitude information from the rover's side? Or are there any documents or tutorails that I can go though to figure out this question? Thanks for the help! Very helpful video!
@@iforce2d I checked the first vdieo you linked here and tried to use your UBX_GPS_SS.ino file to read the NAV_POSLLH message from the C94-M8P rover's side through Arduino. Here are steps that I did: 1. I first enable the NAV-POSLLH message through the u-center software on my PC (disable all other messages). 2. Then, I connected the Arduino Mega pin3 and pin5 with C95-M8P J8 connector with pin9(RX) and pin10(TX). I follow this blog to connect Arduino and C95-M8P, which says on pin9 and pin10 on J8 connector are Rx and Tx. (portal.u-blox.com/s/question/0D52p00008HKD0YCAX/usage-of-uart1-with-c94m8p3-board) 3. I upload the Arduino file on Mega, and open the Serial monitor, but I there is no message from there. Am I doing anything wrong in above steps? Thanks for your help!
Not that I can tell just from reading a comment here. I would suggest using a text message (enable any NMEA message) and just print the incoming stream from the GPS straight to the arduino serial monitor. That would at least let you see if something is coming in.
@@iforce2d Thanks for your reponse! It was actually the buadrate. I change my Arduino baudrate to match the u-center, and it works fine now! Thanks for your help!
Hi, I've got a C94-M8P-3 kit and your video was really helpful for initial setup etc, however I still can't get the modules to output RTCM3 messages. I have followed the quick setup guide, the base module goes into TIME fix mode but then nothing... The RTK LEDs are always off and I can't see any RTCM3 messages on the packet console. I've tried lowering the uart1 baud rate to 9600bps as well as a lot of other options but nothing seems to work. Could you upload the config files somewhere?
Impressive stuff. Totally geeked out. Back around 2000 I was working on a mine truck dispatch system. These two units are doing the same task as 10U rack full of equipment. performance vs cost is outstanding. At $500 I would all over it. How can one get one in Aus ?
Does the GPS module include a compass like most other GPS modules ? If not, due to the strong magnet on the bottom, won't you need to add a compass when using it on your multirotor quads or hex ?
Very impressive, wondering about your application given you're in agriculture. Wonder are there open source projects that build on this data with the ability to traverse a field accurately sowing fertilizer, for example.
I got a question... So far things covered seems to point to this thing being "your-own-GPS-augmentation-network."... Does this thing actually give absolute/relative position information such as heading, pitch, yaw? For the autonomous device, I think the heading from GPS is not going to be there if you are not moving. Can someone knows this stuff also chime in? Thanks.
No, it's a GPS so it gives you lat/lon/alt position. For pitch you would need an accelerometer, and for yaw you would typically use a compass. For heading yes, you could move a bit to and then check which direction you moved in. The difference between this and a regular GPS is that you only need move a few centimeters, and the result is actually reliable - as you can see the result from using a regular GPS to determine movement would only be useful if you're moving fast enough that the error in position becomes insignificant relative to the angle measured (eg. for a plane it's still useful). The RTK module is also accurate enough that you can use two of them on the rover as a 'compass' that does not rely on Earth's magnetic field.
Thanks for the answer iforce2d. I think my mistake was lumping RTK and IMU fusion into just RTK. So, i would still have to use another systems to get the information I need. I have used Bosch BNO055 before. It works very well. I recommend it to you if you are going to need to get any useful kinematic information on the rover you are designing... I think I am confusing myself with the word "kinematics". As in forward/inverse kinematics in robotics: one can get the end effector's position given angles of the joints (all revolute manipulators) using forward, vise versa for the inverse... But in this RTK business, "kinematics" must be the simple velocity and acceleration equations... and doing them in real time, of course. If this is so, then the information I would need should be in the system... Are they not exposing the acceleration information? OR, am I confusing RTK with dead-reckoning this time? Damn it...
Probably would not make ublox very happy, but you should start playing with RTKLIB so I can learn from you :) Supposedly the cheap Neo-6P will work with it.
Got a couple of Neo-M8T coming and going to play around with RTKLIB (Since i'm a cheapskate). Thanks so much for the demo! Got some project inspiration from it.
Nice video! Do you have any suggestions on how to record position in RTK mode? I would like to have e.g. an excel file with coordinates for each second. I am struggling right now with ucenter. Thanks!
You can record the GPS raw output to file like this: ua-cam.com/video/uz8u4VymQiw/v-deo.html To convert it to some other format you would probably need to write your own converter. If I was doing it I would set up the GPS to output UBX messages, then read in those messages as described here: ua-cam.com/video/TwhCX0c8Xe0/v-deo.html ua-cam.com/video/ylxwOg2pXrc/v-deo.html Then write the values to a file as text, with commas between them :)
At what point in the video is that? If you're talking about the empty packet console at 50:40, as the on-screen notation says the next step will show them. Just follow the setup guide steps is all you need to do.
thank you for your answer, now c94 m8u on the way to chile after seeing the accuracy you got. Did you try maximum range of onboard radios? Have you managed to transmit data with other radio?
A little offtop Chris, can You recommend some of your favourite materials about programming ? I mean books or webpages/blogs on which You began and find them valuable. Im aiming in beginner/intermediate level. And what IDE do You use most often ? Arduino ?Eclipse? I would be very grateful for any help man. Best regards.
Since StackOverflow has been around that's probably the site I use most, but that's quite specific to certain problem cases. I don't think I have ever followed a generic 'programming' guide very often, unless to learn the basics of a new API or concept (eg. opengl, glsl) but even then just for a short time at the beginning, then after that just reading the documentation should suffice. The IDE will depend on what you're making - for arduino I use the arduino IDE, for PHP/Javascript I mostly use a text editor, for C/C++ I use QtCreator whenever I can.
You would need to build some kind of logging device, maybe an arduino with an SD card. rtklibexplorer.wordpress.com/2019/10/25/building-a-simple-u-blox-f9p-data-logger-with-a-sparkfun-openlog-board/ Or if you can place a laptop there, maybe use u-center to record to a file. Look for "record mode" in the manual: www.u-blox.com/sites/default/files/u-center_Userguide_%28UBX-13005250%29.pdf
Awesome video, thanks for explanation. Anyone knows how fast you need or should send corection data to "moving" gps ? I would say these data arent changing so much so fast right ?
Depends what the messages are. NAV-PVT should contain everything a flight controller needs, so that would be just one message. But if using other messages, you'll need more than one.
very good. is there any way to dual read GPS and GLONASS double precision floats and add, divide by two to obtain an average? would this yield better results. cheers, sub'd
Hi - I am Engineer working for the NZTA down in Christchurch, I watched this a while back and was recently talking to colleagues about the joys of tracking construction equipment like rollers over relatively short distances (hundreds of metres). Apparently the expensive commercial GPS solutions from roller suppliers are not great and at the price of this solution it’s a no brainer to give it go. We would be happy with +- 1m. Do you think it would work? If so - Would you be interested in some work helping us trialing it?
Assisted GPS uses a base station that could be a very long distance away, seeing different conditions. With this the base station is certain to be giving the best possible result.
Hi I watched your video so interesting , I handled same Neo- M8p RTK . but it had a problem In my module , Rover dosen't fix the FIX mode . so If we move around, we could not get the 2.5cm error range-(chaned float mode ). How could you stay FIX mode when you move your GPS Rover? -We experiment on the outside( open space ) and we chaneged the BASE more good one. but it could not solve the problem Please let me know how to retain the FIX mode .. Thank you.. I'll wait your reply ..
Hello, can you tell me what range it works? The maximum distance that can be between Rover and base that it works.. it would really help me, thanks a lot
In my experience it actually wasn't very good further than about 100 meters or so. That's kinda surprising, so maybe something was wrong with my unit or the antennas...
I know this is a comment on 4 year old video, I just found this video. But, was I the only one that noticed as your rover GPS set was placed 10 meters away the error between the expensive GPS and $20 GPS was a LOT smaller on the cheap GPS? If this is linear, why could a formula be created that can be applied to the Cheap GPS location/distance reading correcting the error, an error correction algorithm?
tbh I am confused by your question. There was nowhere in the demo that the cheap GPS had smaller error. The number on the screens is the distance from where the arduinos were powered up, so when the wooden board was at the starting point the correct answer should have been zero. The expensive one showed 1.4cm and the cheap one showed up to 300cm sometimes.
Sorry, Thought I saw a smaller difference between the two once you moved it to 10 meters away on the fence post. But your videos are great! Thanks. Doesn't look like U-blox will sell a single board. Have to contact them to get the board you are showing. Neo-M8P is around $224 from spark fun. Still not too bad of a price. Thanks for the reply.
can this go with other manufacture brand accesories. if so, can it connect with with freewave radio module from swift navigation. Also what is the longest distance you went besides 10 meters.
The connection between the base station and rover requires a serial modem, I think any would do. I went a few hundred meters away before the connection started to get patchy (both ends at ground level, some trees in the way).
How is the accuracy of c94-m8p-2 compared to neo-m8p-2? Also, is there any way to automatically enter "fix-mode" again after crossing area with bad signal?
Is it possible to receive NMEA from rover and base from only USB port? So far I have some unknown messages on USB and I think it's from rover but cannot read message. Any advice?
My rover sends back his position to base and then I can read it but I haven't figured it out how to read it by one interface. I have to use USB and RS232 to read both. I'd like to read both by USB and wonder if you know if is this possible.
What am I missing in survey-in? It took several day but still with 5cm stdev? And RTK just can't be precise. Anyone has some advice on survey-in process?
@@iforce2d Thanks for the reply! But how did you setup your base station coordinate to get the cm level distance in "Simple test/demo vs M8N"? Just set 1m in TMODE3 is enough? Or I have to find the actual position of base station (cm level) in other way?
6 років тому
Damn... but dude, if the precision is in the centimeter or milimeter range, there is no error you see there. The earth's crust moves like that from tides and tectonics. :) Well, this integrates accelerometers with GPS, right?
sorry ...distance measure equipment... found and looked at the link but I'm not literate in code, but maybe I can figure out how to install your binary
tab dougherty I was wondering the same thing. I gather it’s both for 399. When you click ‘kit includes’ it says 2 x application boards(with gps) / cables etc
Hey, I've searched for you on Patreon with no luck. I find your channel very inspiring and would love to contribute (modestly). Is there a way I can do this?
I might make a patreon someday. For now if you buy something on Banggood you can use my affiliate link, I'll get some commission. Add this to the product URL: ?p=2K090270884220140865
@@iforce2d thank you very much for your help!! One last question. Does the base station need a microcontroller? Or it just need just power and configuration?
I see you use mbed IDE to program the microcontroller. I guess it's a STM32 µc. What made you decide to choose mbed instead of arduino Ide for stm32. There is eclipse as well for arduino that you can use to program stm32 ...
Skipped over most of the second half but I can see this being most useful in mapping small areas, farms for example, fertiliser load placement and the like. Don't see that kind of accuracy being important to the hobby flyer.
@@iforce2d Thanks for your answer. Sorry, I confused the STM32 with an Arduino, I just checked the video. I'm trying to do your project, I want to use this for hillside monitoring, but I still can't get even 4 meters of precision, maybe because of my location (Colombia).
This looks very promising; nice work. I live in Auckland, want to tow a cruising yacht with a kite and am looking for a (slow) moving base system to locate the kite's (rover) position in the sky (50m away) relative to the yacht, so as to provide info for the kite autopilot. (skysails-yacht.com) I need two-way comms for this, as the autopilot is based on the yacht, unless I somehow decide that the kite/gondola is the base and the yacht is the rover. Would be interested in your thoughts?
If the relative distance is 50m or so, you don't need a fancy RTK unit like the one in this video. Regular GPS are accurate to about 2-3 meters which is enough to relate direction between two things to each other pretty well, eg. ua-cam.com/video/uilJtMOip5k/v-deo.html If I'm understanding the requirements correctly, one-way comm should be enough, that is, yacht sends its location and the desired relative angle. Yacht doesn't need to know where the kite is I would think. hmm.... unless the autopilot handles tacking and upwind travel etc. One of those radio modems as used for Ardupilot etc would work. I don't know much about sailing but I thought sailing boats need massive keels purposely made to resist the sideways force of a tack configuration. I can't imagine how it would be possible to tack with the kite either, or how much fuel would be saved overall by zigzagging to accommodate the sail. The site you mention says they have all the autopilot stuff figured out, are you gonna make your own?
@@iforce2d Thanks very much for the thoughtful reply. I do have a few GPS units and even though their horizontal position is known quite well as you say, the vertical accuracy is poor, and the Skysails autopilot, which I will be using, needs to know an accurate height, especially since it will often be flying the kite quite close to the water surface. (I now have this ublox unit on order, and with a little help from my techo friends, and your excellent video, hope to get it working OK). The autopilot also uses other information like line tension, wind speed and boat speed.. The autopilot knows the relative location by measuring two angles of the tow line at the towing block, and the distance to the gondola, but this is heavy, expensive and clumsy and doesn't allow us to move the tow point around easily, which you would often want to do on a yacht. So I am keen to give the RTK a go - it will be a big improvement if we can get a reliable relative position this way. Most of the time, the kite is like a normal sail, except up in the air a bit and not trying to tip the boat over. For gybing the yacht (changing course from one direction to the other with the wind behind) we would normally let the kite fly up to the top (12 o'clock) during the manoeuvre, and then fly down to the best pulling angle once we are settled on the new course. We can also hand-fly the kite when needed just by looking at it and pressing steering keys on the controller computer. If you want to know more, we can always take this conversation off line, just email me at tom.schnack@gmail.com
Perhaps I didn't point out clearly enough in the video, but RTK works by having a stationary base station, that is known to be not moving. Any inaccuracies caused by atmospheric conditions will then also be known, and 'subtracted' from the position of other (moving) modules in the same vicinity. I really don't think it will work properly when the base station moves. If sensing altitude accurately is the main concern, I think you might be better off with a well shielded barometer, which will be more accurate than any kind of GPS. A similar barometer on the boat that never changes altitude significantly could be used to adjust for long-term drift due to changes in weather conditions, ie. it would perform a similar function to the GPS 'base station'. I thought the Skysails system had this all figured out though?
the blue case is to stop the TCXO (Temp compensated oscillator) from drifting due to air flow over it
Another great video you are so natural it explaining this technical subject and make it not only interesting but comprehensive
Excellent video for a newbie getting into RTK's such as myself. Wish I had your tutorial 4 months ago when I started using the C94 M8P. I like how you explain the small details which the guide does not cover, and those are really important for people like me that do not really know whats going on, but really would like to see some results. Looking forward for more videos.
Small correction: In case of GPS, each satellite has a completely different PRN code (pseudo random number). If they transmitted the same PRN then you'd have trouble picking up different ranges (if dopplers were sufficiently close together) due to cross-correlation. GPS uses CDMA so that is the case. This is true though in case of GLONASS which uses FDMA and where all satellites do in fact transmit the same PRN (but on different frequency bands).
EDIT: You may also be able to get network RTK corrections (not sure if available where you are). For that you'd need a single RTK enabled rover (no base required) + a cell modem or something to access the internet (even a cell phone). The corrections are broadcast over the internet (generally using NTRIP) and are fed into your receiver which replaces the base. Network RTK in most case will have extended range when compared to single base RTK (sometimes up to 50km from nearest base).
EDIT2: FLOAT ambiguity simply refers to the accuracy of the ambiguity resolution. i.e. for satellite X the ambiguity was resolved to be 3.45 cycles which has a floating point resolution and hence the name. Fixed refers to fixed or integer fixed ambiguity and it is more precise; this means that the ambiguities are resolved and fixed to actual whole integers. For most applications (less than decimeter) FLOAT is good enough however you will see a quantum leap when going from float to fixed in terms of stdev. Generally on start up you'll see the rover start with FLOAT and then step to FIXED.
uBlox is an amazing company. I brought a M8U from them (but its only about 70$), but I screwed up the silk screen (on my designed PCB) and mixed up Vcc and GND.... They sent me a new one and wished me good luck with my project :)
whats difference between m8u and m8n? just wondering if i should buy m8u cheers
There's a nice overview here: www.u-blox.com/sites/default/files/products/documents/GNSS-module-selector_Overview_%28UBX-14000426%29.pdf The M8U includes an IMU
Thanks for that amazing video - and it's *not* to long - great work!
greetings from the opposite side of the planet.
Oh that looks very nice indeed. I guess it would improve the autonomous farm rover application a lot. I'd also be curios to see a multicopter using this and doing a position hold, that should be awesome!
Thanks for another very interesting video
Like a true surveyor, I shall attempt to follow in your footsteps, thanks.
I laughed every time you said "peasants gps". Thanks for a very educational video.
same, notably "peasants" nowadays use machines with RTK for their peasantry work
Thanks for the explanation of RTK. Very interesting technology.
Fantastic, you are my favorite teacher of all time. Thanks for sharing.
Your doing a better video than ublox webinar
At that price point, you would have thought a Polysilicon fuse and transorb would have been in the budget for them to include on the modules!
Also, you may wish to locate and purchase some RS422 transceivers for doing serial communications over such long cables. They'll be much more reliable in the presence of interference.
Great video!
Yeah, I was worried about that so I compared the input and output signals with my scope. I had to go all the way to 100ns/div to see any difference, so I figure this particular link and cable combination should be safe for a pretty good distance: www.iforce2d.net/tmp/MAP004.png
iforce2d for your setup purposes it’ll probably be just fine, but a car with a noise alternator driving by, a farmer with a walkie talkie or perhaps a nearby electrical storm could corrupt the occasional packet.
I guess eventually it’ll be standalone (solar powered?) and do all its work over the onboard telemetry link...
Excellent intel, much appreciated!
I'll try to make a base/rover setup for an autonomous lawn mower with obsolete u-blox LEA-5T modules, which can be bought used for less than 20 euros on eBay.
The speed of light is 300,000Km/s, and not 300,000 Km/h (as shown at 8:14 in the video). But anyway, you did a great job.
That's awesome.
Great video... thanks, BTW RTK built-in mission planner... easy setup, been using c94 as the base and drotec XL as the rover
Great video. I would like to hear about the range you are getting from your base station corrections and how it works around objects.
Jeez I could use that for so many things :) very impressive.
Very educational.
Dumb question. How much accuracy would you expect to get if you had a DIY differential system with (2) M8Ns? (1) base station M8N set on your home's roof or other set location and (1) rover M8N with the base sending offset/correction info via nRF24L01? I would think that the base would send info like direction and distance of the offset.
Good question, I might try it sometime.
It's more than just a differential M8N actually. What I think the RTK does is that the base station averages the clock phase in order to extract a very accurate clock data stream of its own. The GPS/GLONASS satellites each have their own Rubidium reference clock from which the data streams are generated. Regular GPS modules derive this clock stream during fix computation from received signals (what is indicated by the "3D"), making it subject to any path variations. I suspect the base station stops doing this once it has established it's own version of the clock and switches to using this (hence the "TIME" indicator). For this to work, the base station needs really good time keeping of its own - maybe having a Rubidium clock, for instance. In the case of the unit being tested, I suspect that's what the blue plastic box is all about - temperature regulation for a stable local clock.
Using an M8N as the base station does not give this clock/temperature stability. Assuming the hardware of the M8N and M8P is the same, regulating the temperature of the M8N GPS unit might do the trick. The blue box encloses the GPS unit with it's crystal clock. It excludes the RF transceiver module because that usually sees a lot of temperature variation from the output amplifier.
@Daniel Wee, If your theory is correct... I bet the offset would be as slow moving as the temperature changes, and relativistic anyway. This probably means that for your average 20-minute mission, it won't matter; RTL will still be centimeter-accurate. Ardupilot (and DJI) is mostly all relativistic navigation, cause you're lucky if it lines up with google maps at
electronics.stackexchange.com/questions/234038/gps-accuracy-is-it-relative-or-absolute blog.dronedeploy.com/accuracy-in-drone-mapping-what-you-need-to-know-10322d8512bb
I'm surprised to hear that you got a 100m difference in subsequent flights to the same place. You mean when you launched from a different location the second time right? I don't know how other firmwares do it, but in the iNav source code I have seen that it does most of the position calculation in a meters-based local coordinate system with the launch point as the origin. So in theory when you give it a lat/lon target, the location it derives may not be quite the same as when starting from a different origin, but 100m seems pretty outrageous. With a M8N I have rarely seen it more than about 3 meters off the google map image. Even if you doubled that and assumed each origin was off in the worst possible direction, that's still only about 12 meters discrepancy between origins. The math should be using double precision so even at 2 miles away from the origin I would expect a lot better than 100m difference.
So if the difference is due to the inner workings of the flight controller math then RTK will not help much. But if it's not, then RTK would work great - just don't move the base station of course.
Great stuff again Chris.
The nail and MCU shows why you can never have enough breadboards!
I was about to buy a few more from BG but hoped to find a wider (not longer) plastic breadboard with no luck. Anyone know of where to find them?
BTW, I think "Hammond" is the name of a large enclosure manufacturer.
hi! great video. I´m a total station surveyor trying start in the gps world. Now it's getting much cheaper and more accurate. Not very convinced by the reach rs solution, maybe i´l dedicate some time to a diy solution. Can you talk a little bit more about your microcontroler solution? did you get to program any datum conversion algorithms , or just simple angle to x,y calculation?
btw, love the sound of the birds and other animals on the background!! you are definitelly a fortunate man!
The numbers shown are just taken from the NAV-PVT message, which has lat/lon/hdop etc included. The program is basically the same as what I looked at in detail here, but using a different message type: ua-cam.com/video/TwhCX0c8Xe0/v-deo.html
Thank you for the video. Did you write the C code for ubxDistanceMeter and how did you learn how to get the GPS coords from the USB/UART port please?
ua-cam.com/video/TwhCX0c8Xe0/v-deo.html
Hi, I would like to ask a question.
I see a very good precision in the final result. Do you use the position calibration required by the Base Station in survey-in mode or fixed-mode?
Thank's
All the results I showed were after doing the survey-in, but I think you can still use the RTK feature even while the survey-in is in progress.
Great video! Thanks for the instructions!
Is there a way that I can use an Arduino board to read lattitude and longtitude information from the rover's side?
Or are there any documents or tutorails that I can go though to figure out this question?
Thanks for the help! Very helpful video!
ua-cam.com/video/TwhCX0c8Xe0/v-deo.html
ua-cam.com/video/ylxwOg2pXrc/v-deo.html
@@iforce2d Thanks a lot!
@@iforce2d I checked the first vdieo you linked here and tried to use your UBX_GPS_SS.ino file to read the NAV_POSLLH message from the C94-M8P rover's side through Arduino.
Here are steps that I did:
1. I first enable the NAV-POSLLH message through the u-center software on my PC (disable all other messages).
2. Then, I connected the Arduino Mega pin3 and pin5 with C95-M8P J8 connector with pin9(RX) and pin10(TX). I follow this blog to connect Arduino and C95-M8P, which says on pin9 and pin10 on J8 connector are Rx and Tx. (portal.u-blox.com/s/question/0D52p00008HKD0YCAX/usage-of-uart1-with-c94m8p3-board)
3. I upload the Arduino file on Mega, and open the Serial monitor, but I there is no message from there.
Am I doing anything wrong in above steps?
Thanks for your help!
Not that I can tell just from reading a comment here. I would suggest using a text message (enable any NMEA message) and just print the incoming stream from the GPS straight to the arduino serial monitor. That would at least let you see if something is coming in.
@@iforce2d Thanks for your reponse! It was actually the buadrate. I change my Arduino baudrate to match the u-center, and it works fine now! Thanks for your help!
Hi, I've got a C94-M8P-3 kit and your video was really helpful for initial setup etc, however I still can't get the modules to output RTCM3 messages.
I have followed the quick setup guide, the base module goes into TIME fix mode but then nothing... The RTK LEDs are always off and I can't see any RTCM3 messages on the packet console. I've tried lowering the uart1 baud rate to 9600bps as well as a lot of other options but nothing seems to work.
Could you upload the config files somewhere?
31:09... Yeah I agree... It may cost $400 USD if you're not careful... ;)
Been there... Done that...
lasersbee Yes, but as Daniel Wee mentioned in his comment it might be just fried regulator and radio and ublox could be just fine.
Also be careful with the micro USB - I ripped it off accidentally. Fortunately my company paid for it XD
Impressive stuff. Totally geeked out. Back around 2000 I was working on a mine truck dispatch system. These two units are doing the same task as 10U rack full of equipment.
performance vs cost is outstanding. At $500 I would all over it. How can one get one in Aus ?
Link is in the description.
nice explanations there.
Does the GPS module include a compass like most other GPS modules ? If not, due to the strong magnet on the bottom, won't you need to add a compass when using it on your multirotor quads or hex ?
No. R/C planes/drones etc are not really the target market.
Very impressive, wondering about your application given you're in agriculture. Wonder are there open source projects that build on this data with the ability to traverse a field accurately sowing fertilizer, for example.
There are some compagnies that uses gps to drive a feeding installation for cows. ua-cam.com/video/Uw-pVl1Y45o/v-deo.html for example.
I got a question... So far things covered seems to point to this thing being "your-own-GPS-augmentation-network."... Does this thing actually give absolute/relative position information such as heading, pitch, yaw? For the autonomous device, I think the heading from GPS is not going to be there if you are not moving. Can someone knows this stuff also chime in? Thanks.
No, it's a GPS so it gives you lat/lon/alt position. For pitch you would need an accelerometer, and for yaw you would typically use a compass. For heading yes, you could move a bit to and then check which direction you moved in. The difference between this and a regular GPS is that you only need move a few centimeters, and the result is actually reliable - as you can see the result from using a regular GPS to determine movement would only be useful if you're moving fast enough that the error in position becomes insignificant relative to the angle measured (eg. for a plane it's still useful). The RTK module is also accurate enough that you can use two of them on the rover as a 'compass' that does not rely on Earth's magnetic field.
Thanks for the answer iforce2d. I think my mistake was lumping RTK and IMU fusion into just RTK. So, i would still have to use another systems to get the information I need. I have used Bosch BNO055 before. It works very well. I recommend it to you if you are going to need to get any useful kinematic information on the rover you are designing...
I think I am confusing myself with the word "kinematics". As in forward/inverse kinematics in robotics: one can get the end effector's position given angles of the joints (all revolute manipulators) using forward, vise versa for the inverse... But in this RTK business, "kinematics" must be the simple velocity and acceleration equations... and doing them in real time, of course. If this is so, then the information I would need should be in the system... Are they not exposing the acceleration information? OR, am I confusing RTK with dead-reckoning this time? Damn it...
Hi, Yes.. the rover side GPS and the controller have mags built-in
Ok then. Thanks.
Probably would not make ublox very happy, but you should start playing with RTKLIB so I can learn from you :) Supposedly the cheap Neo-6P will work with it.
Got a couple of Neo-M8T coming and going to play around with RTKLIB (Since i'm a cheapskate). Thanks so much for the demo! Got some project inspiration from it.
hey sory cold u pleas hepl me out i got the c94 bords but cant get neeer even that hig acurecy like you got, can ve get in tuch?
Nice video! Do you have any suggestions on how to record position in RTK mode? I would like to have e.g. an excel file with coordinates for each second. I am struggling right now with ucenter. Thanks!
You can record the GPS raw output to file like this:
ua-cam.com/video/uz8u4VymQiw/v-deo.html
To convert it to some other format you would probably need to write your own converter. If I was doing it I would set up the GPS to output UBX messages, then read in those messages as described here:
ua-cam.com/video/TwhCX0c8Xe0/v-deo.html
ua-cam.com/video/ylxwOg2pXrc/v-deo.html
Then write the values to a file as text, with commas between them :)
Good work here. Are you able to make a kit for me? I farm in the rift valley region of Kenya
Pretty nice
Thanks for the video - you say you can make changes so that you can see RTCM3 messages in packet window .... But what are those changes please?
At what point in the video is that? If you're talking about the empty packet console at 50:40, as the on-screen notation says the next step will show them. Just follow the setup guide steps is all you need to do.
thank you for your answer,
now c94 m8u on the way to chile after seeing the accuracy you got. Did you try maximum range of onboard radios? Have you managed to transmit data with other radio?
A little offtop Chris, can You recommend some of your favourite materials about programming ? I mean books or webpages/blogs on which You began and find them valuable. Im aiming in beginner/intermediate level. And what IDE do You use most often ? Arduino ?Eclipse? I would be very grateful for any help man.
Best regards.
Since StackOverflow has been around that's probably the site I use most, but that's quite specific to certain problem cases. I don't think I have ever followed a generic 'programming' guide very often, unless to learn the basics of a new API or concept (eg. opengl, glsl) but even then just for a short time at the beginning, then after that just reading the documentation should suffice.
The IDE will depend on what you're making - for arduino I use the arduino IDE, for PHP/Javascript I mostly use a text editor, for C/C++ I use QtCreator whenever I can.
Thanks!
hello, great video! please, how do I save both the rover and base station UBX data for post-processing?
You would need to build some kind of logging device, maybe an arduino with an SD card.
rtklibexplorer.wordpress.com/2019/10/25/building-a-simple-u-blox-f9p-data-logger-with-a-sparkfun-openlog-board/
Or if you can place a laptop there, maybe use u-center to record to a file. Look for "record mode" in the manual:
www.u-blox.com/sites/default/files/u-center_Userguide_%28UBX-13005250%29.pdf
Can this be used for indoor tracking? Is there a significant performance difference?
I'm not aware of any GPS receiver that would work indoors with any kind of usable accuracy. Unless maybe the roof is made of plastic :)
Do we need to connect both rover and base to same laptop?
Awesome video, thanks for explanation.
Anyone knows how fast you need or should send corection data to "moving" gps ?
I would say these data arent changing so much so fast right ?
Yeah, I would expect so too. But I'm not certain.
@@iforce2d Is possible to log these data or something? 🙂
To chceck how often they change and how much.
Hello again Master kiwi 😄how many of those messages do I need active for an average flight controller, an f4 to be exact and its just a m8 gps no rtk
Depends what the messages are. NAV-PVT should contain everything a flight controller needs, so that would be just one message. But if using other messages, you'll need more than one.
very good. is there any way to dual read GPS and GLONASS double precision floats and add, divide by two to obtain an average? would this yield better results. cheers, sub'd
thanks for your sharing , how can i calculate mm unit data, like 56:52?
Source code is linked to in the description...
Hi - I am Engineer working for the NZTA down in Christchurch, I watched this a while back and was recently talking to colleagues about the joys of tracking construction equipment like rollers over relatively short distances (hundreds of metres). Apparently the expensive commercial GPS solutions from roller suppliers are not great and at the price of this solution it’s a no brainer to give it go. We would be happy with +- 1m. Do you think it would work? If so - Would you be interested in some work helping us trialing it?
Hey could you make a video about your own RF receiver and transmitter? And maybe explain everything ;p
Not sure what you mean by my own rx and tx.... is it related to this video or do you mean about nRF24L01 stuff?
So whats difference A-GPS which we know it used by smartphone and the RTK?
Assisted GPS uses a base station that could be a very long distance away, seeing different conditions. With this the base station is certain to be giving the best possible result.
Hi I watched your video so interesting , I handled same Neo- M8p RTK . but it had a problem In my module , Rover dosen't fix the FIX mode . so If we move around, we could not get the 2.5cm error range-(chaned float mode ). How could you stay FIX mode when you move your GPS Rover? -We experiment on the outside( open space ) and we chaneged the BASE more good one. but it could not solve the problem Please let me know how to retain the FIX mode .. Thank you.. I'll wait your reply ..
It should just retain the fix if the signal reception is good enough. Maybe your rover is generating some interference?
iforce2d okay .. thank you I’ll try once again
Hello, can you tell me what range it works? The maximum distance that can be between Rover and base that it works.. it would really help me, thanks a lot
In my experience it actually wasn't very good further than about 100 meters or so. That's kinda surprising, so maybe something was wrong with my unit or the antennas...
like the differential gnss explanation. help me understand in a second.
I know this is a comment on 4 year old video, I just found this video.
But, was I the only one that noticed as your rover GPS set was placed 10 meters away the error between the expensive GPS and $20 GPS was a LOT smaller on the cheap GPS? If this is linear, why could a formula be created that can be applied to the Cheap GPS location/distance reading correcting the error, an error correction algorithm?
tbh I am confused by your question. There was nowhere in the demo that the cheap GPS had smaller error. The number on the screens is the distance from where the arduinos were powered up, so when the wooden board was at the starting point the correct answer should have been zero. The expensive one showed 1.4cm and the cheap one showed up to 300cm sometimes.
Sorry, Thought I saw a smaller difference between the two once you moved it to 10 meters away on the fence post. But your videos are great! Thanks. Doesn't look like U-blox will sell a single board. Have to contact them to get the board you are showing. Neo-M8P is around $224 from spark fun. Still not too bad of a price.
Thanks for the reply.
Would it be possible to do a video about interfacing these with MCUs?
ua-cam.com/video/TwhCX0c8Xe0/v-deo.html
ua-cam.com/video/ylxwOg2pXrc/v-deo.html
can this go with other manufacture brand accesories. if so, can it connect with with freewave radio module from swift navigation. Also what is the longest distance you went besides 10 meters.
The connection between the base station and rover requires a serial modem, I think any would do. I went a few hundred meters away before the connection started to get patchy (both ends at ground level, some trees in the way).
How is the accuracy of c94-m8p-2 compared to neo-m8p-2? Also, is there any way to automatically enter "fix-mode" again after crossing area with bad signal?
I don't have the neo-m8p to compare. The module will always use the best mode it can.
iforce2d what’s the best accuracy you get? Also, what’s the delay if you use it for real-time positioning of a car?
Hey there!
What are actually these small two displays, you use in the beginning?! How are they called?
search for "0.96 oled screen" many places sell them
Is it possible to receive NMEA from rover and base from only USB port? So far I have some unknown messages on USB and I think it's from rover but cannot read message. Any advice?
There are many different ways to set up the message passing, but if you do it the standard way like I did, the rover never sends any messages at all.
My rover sends back his position to base and then I can read it but I haven't figured it out how to read it by one interface. I have to use USB and RS232 to read both. I'd like to read both by USB and wonder if you know if is this possible.
Sir now am student i can not make ublox with can now get data but can not talk to server . could you please help us.
What am I missing in survey-in? It took several day but still with 5cm stdev?
And RTK just can't be precise. Anyone has some advice on survey-in process?
40:20 Try 1 meter as the target. These GPS are not capable of 5cm absolute accuracy.
@@iforce2d Thanks for the reply!
But how did you setup your base station coordinate to get the cm level distance in "Simple test/demo vs M8N"?
Just set 1m in TMODE3 is enough? Or I have to find the actual position of base station (cm level) in other way?
Damn... but dude, if the precision is in the centimeter or milimeter range, there is no error you see there. The earth's crust moves like that from tides and tectonics. :)
Well, this integrates accelerometers with GPS, right?
iforce2d any chance of getting a wiring diagram of your DME looks very handy
Not sure what you mean by DME. Link to the code is in the description.
sorry ...distance measure equipment... found and looked at the link but I'm not literate in code, but maybe I can figure out how to install your binary
So 399.00 gets you the base and rover or are they 400 each sry if you mentioned it i did not hear 800 or 400?
tab dougherty I was wondering the same thing. I gather it’s both for 399. When you click ‘kit includes’ it says 2 x application boards(with gps) / cables etc
Hey, I've searched for you on Patreon with no luck. I find your channel very inspiring and would love to contribute (modestly). Is there a way I can do this?
I might make a patreon someday. For now if you buy something on Banggood you can use my affiliate link, I'll get some commission. Add this to the product URL: ?p=2K090270884220140865
Where can i find the Arduino code for rover and base station?
I used mbed for this, not arduino.
@@iforce2d mbed is a microcontroller too, right? so, mbed needs code to compile for the RTK. can i find this code somewhere?
It's in the video description....
mbed is a method of creating code for various microcontrollers:
ua-cam.com/video/Er46sIHniO4/v-deo.html
@@iforce2d thank you very much for your help!! One last question. Does the base station need a microcontroller? Or it just need just power and configuration?
He setted the base station board accuracy to one meter but at the and of the project he get cm level presicion how is it possible?
The cm accuracy is relative to the base station, not global position.
what kind of radio module does it cost?
This is a development board with the radio included...
I see you use mbed IDE to program the microcontroller. I guess it's a STM32 µc. What made you decide to choose mbed instead of arduino Ide for stm32. There is eclipse as well for arduino that you can use to program stm32 ...
Skipped over most of the second half but I can see this being most useful in mapping small areas, farms for example, fertiliser load placement and the like. Don't see that kind of accuracy being important to the hobby flyer.
Hammond is just the manufacturer of the plastic case. The rest is the product number.
31:11 hmm that's how u damaged the previous module
Can you share the Arduino code?
There is no arduino code, this is using mbed, link is in the description.
@@iforce2d Thanks for your answer. Sorry, I confused the STM32 with an Arduino, I just checked the video. I'm trying to do your project, I want to use this for hillside monitoring, but I still can't get even 4 meters of precision, maybe because of my location (Colombia).
Nevermind LOL found it
This looks very promising; nice work. I live in Auckland, want to tow a cruising yacht with a kite and am looking for a (slow) moving base system to locate the kite's (rover) position in the sky (50m away) relative to the yacht, so as to provide info for the kite autopilot. (skysails-yacht.com) I need two-way comms for this, as the autopilot is based on the yacht, unless I somehow decide that the kite/gondola is the base and the yacht is the rover. Would be interested in your thoughts?
If the relative distance is 50m or so, you don't need a fancy RTK unit like the one in this video. Regular GPS are accurate to about 2-3 meters which is enough to relate direction between two things to each other pretty well, eg. ua-cam.com/video/uilJtMOip5k/v-deo.html
If I'm understanding the requirements correctly, one-way comm should be enough, that is, yacht sends its location and the desired relative angle. Yacht doesn't need to know where the kite is I would think. hmm.... unless the autopilot handles tacking and upwind travel etc. One of those radio modems as used for Ardupilot etc would work. I don't know much about sailing but I thought sailing boats need massive keels purposely made to resist the sideways force of a tack configuration. I can't imagine how it would be possible to tack with the kite either, or how much fuel would be saved overall by zigzagging to accommodate the sail. The site you mention says they have all the autopilot stuff figured out, are you gonna make your own?
@@iforce2d Thanks very much for the thoughtful reply. I do have a few GPS units and even though their horizontal position is known quite well as you say, the vertical accuracy is poor, and the Skysails autopilot, which I will be using, needs to know an accurate height, especially since it will often be flying the kite quite close to the water surface. (I now have this ublox unit on order, and with a little help from my techo friends, and your excellent video, hope to get it working OK). The autopilot also uses other information like line tension, wind speed and boat speed.. The autopilot knows the relative location by measuring two angles of the tow line at the towing block, and the distance to the gondola, but this is heavy, expensive and clumsy and doesn't allow us to move the tow point around easily, which you would often want to do on a yacht. So I am keen to give the RTK a go - it will be a big improvement if we can get a reliable relative position this way. Most of the time, the kite is like a normal sail, except up in the air a bit and not trying to tip the boat over. For gybing the yacht (changing course from one direction to the other with the wind behind) we would normally let the kite fly up to the top (12 o'clock) during the manoeuvre, and then fly down to the best pulling angle once we are settled on the new course. We can also hand-fly the kite when needed just by looking at it and pressing steering keys on the controller computer. If you want to know more, we can always take this conversation off line, just email me at tom.schnack@gmail.com
Perhaps I didn't point out clearly enough in the video, but RTK works by having a stationary base station, that is known to be not moving. Any inaccuracies caused by atmospheric conditions will then also be known, and 'subtracted' from the position of other (moving) modules in the same vicinity. I really don't think it will work properly when the base station moves.
If sensing altitude accurately is the main concern, I think you might be better off with a well shielded barometer, which will be more accurate than any kind of GPS. A similar barometer on the boat that never changes altitude significantly could be used to adjust for long-term drift due to changes in weather conditions, ie. it would perform a similar function to the GPS 'base station'.
I thought the Skysails system had this all figured out though?
@@iforce2d moving base is not common, but not unknown. www.u-blox.com/sites/default/files/ZED-F9P-MovingBase_AppNote_%28UBX-19009093%29.pdf
Thanks for it... anyone of you have already used this gps with ARDUINO?
Por favor, agregar subtítulos en español!!!
i can't disable nmea child messsge.
If anyone wants configuration files, I've got some here: hackaday.io/project/90057/instructions
Peasants GPS Haha