Total newbie with lots of dumb questions

OK, I have an M600P, Ronin MX gimbal, Sony A7r4 61 MP for developing Ai in the ag industry. We are building Ag intelligence that seeks specifics anomalies in hemp. This is a non typical setup so flight planners and hardware has been difficult.
I purchased the M2 with antenna and hot shoe adapter to synch locations with the shot. This camera does not even come with gps, so up to this point I have been strapping my cell phone to the aircraft and blue tooth connect to the camera. It stamped each pic pretty good.
Ok, now I’m required to go RTK with <5cm accuracy. I have everything mounted on the aircraft quite well, powered up, it’s getting good signal, I uploaded the emlid app on my phone and it communicates good as well. So now I have to mash a bunch of RTK info into my small brain and it hurts.
I went to RTK2GO and registered for an NTRIP, there it tells me I need a SNAP account as well. Ok, did it. Lots of confusing instructions with links to more confusing instructions. I think I can plan a flight and capture stamped images.
I’ve been using just the DJI GS PRO for flight planning and it stinks. I think I’ll go back to Pix4D, anyway, with this camera, it has not been flight planner friendly as it’s third party hardware. So I guesstimate airspeed, overlaps, altitude and set the trigger to every 2 seconds.
Ok, dumb questions. Do I really need and RS2 or similar expensive ground control if I can connect to an NTRIP?
Can I just capture a static gcp as reference and still get the accuracy I require. If so, I’m not sure how to do that either.
Where can I find some clear step by step instructions from power up to uploading RTK files?
Are the day’s RTK files stored on the M2 for access later?
At the end of each day, I’m to upload time and location stamped images, and RTK files to the processing dashboard.
I’m finding way too much info to sift through to find just the nuggets I actually need. Not too many forum topics cover enough info I need.
And if this is way too common and dumb for y’all, I’ll go stand against that wall over there for the firing squad. Any and all info is greatly appreciated.
Thank you.

Hi Dave! Welcome and hopefully we can walk you through a good start. First question is what is it that makes you think that you need <5cm accuracy? Are you trying to do survey mapping? Didn’t sound like it from your introduction. Are you navigating in tight spaces? The M2 setup you mention will not help navigation.

Flight planning with any custom setup is painful. Have you looked at DroneDeploy? You would still need to set image capture on an internal timer, but they can control the M600 as well and their flight planning now allows you to pic a camera for which you are using. All you would have to do is find a camera on the list that has the same FOV and the flight planning is done.

This really depends on how you answer my first two questions. Initially I am going to say yes you should get an RS2. Otherwise you are going to run into the scenario that MANY Phantom 4 RTK users are running into being that NTRIPs,

  1. Require a subscription and account and all services are not the same which means if you cover a large area you may need more than one account.
  2. Are not always equal. They can vary in performance, but also in the types of corrections they provide which is very important when tagging the images directly.
  3. Are not always close enough which means that accuracies can suffer and RTK solutions can become erratic.

Going to the level of a solution that I think you are working towards and the accuracies that you are requiring I don’t see how an RS2 receiver for use as a base station is considered that expensive considering the drone cost 2.5x the cost of it. You might also think about using another M2 as a base, but that will depend on how far you will be flying from the base. I think it will be fine as long as you are abiding by VLOS regulations, but the fact that you are stating AI makes me thinking your solution may be going beyond VLOS. That all said having a local base station is the most reliable and flexible solution for site mapping and if the solution is eventually packaged to individuals (growers) then will each of them be expected to keep up with NTRIP accounts? Or would it be easier for them to have a permanent onsite base station?

I believe with that camera it may be possible to geotag the images directly, but if not look at Geosetter and the associated threads on this forum and forum.dronedeploy.com.

Yes, but you are getting into territory that the thinking may change depending on the answers to the questions above, but also what happens if you are not able to tag the images directly. Then PPK might be a better solution.

Looking forward to learning more about what you are trying to do!

3 Likes

Let’s see, I have used Pix4D before and really liked the simplicity and reliability I had with the Phantom4 system. DP has some mixed reviews by persons in this project.
<5cm accuracy as required by the software developer to allow cultivators in the field to see the targeted anomalies, after processing, and displayed on their phone app to walk directly to a pinpointed object for eradication. This camera does not geo tag images, as described above, I used my cell phone strapped to the aircraft to blue tooth gps stamps to images, and it worked pretty good, but we require more accuracy.
The cultivators current method of visual inspections is hiring a full time team of field hands to walk each row and visually inspect. They do this daily as by the time they reach the end of a field, they start again at the beginning. This is quite costly with some risk involved to manual labor. I myself can do the same acreage in one day, and in theory, return results within 48 hours.
At this R&D stage, I am covering the same 75 acres every week, so it’s all within VLOS. And I expect most in the future will be as well. The longest side of this plot is 1600 feet (I believe).
And yes, I’m grasping at straws at this time as the cost of all this equipment has put our account in the “need only” category. It’s a work for work deal with zero income at this time. Our cost so far is well above the 2.5 times, as extra battery sets, and other support hardware puts us in the nice used car range. And we still need two more sets of six batteries with charger to stay aloft more time than charging batteries. That’s an additional $3500 (ouch!).
But we are sacrificing all, for next grow season we will the master licensing distributor in the US. So I’m not ready to sell the fishing boat yet, but this opportunity can not be passed.
Perhaps I could use an NTRIP service for R&D purposes, just to get close and acquire the GS2 after we get funding. We have one investor on deck, but I have to become fluent, or at least build a solid workflow system from which to learn and deliver. I need to deliver every week and get past my learning curve which resembles a dry lake bed.
I think I can get away with NTRIP accuracy until funding and process becomes better, but for now, I am struggling with affording this project and learning best practices. If I can just upload a complete package for now, fine accuracy will come later. The cultivators are not paying or relying on our services at this time, but allowing us to use their crop as a test bed for future price discounts once we go to market.
Oh, and I’m located in Sutter County California, that’s north central valley, just above Yuba City

To each their own, but I have also heard the opposite and in my opinion from use Pix4D Cloud is subpar and Pix4D Desktop is way more complicated and resource hungry than it should be. DroneDeploy is the simplest all-inclusive cloud solution there is that I have tested and their functionality is growing by leaps and bounds every day. They have a very strong Ag following. Their problem right now is that they have invested so much effort in the project management that the huge releases in flight have become buggy. We are working through them one by one and version v4.12 for iOS is now B+ stable. I do fly with Pix4D capture if anything really seems wrong in the field with DroneDeploy, but their app has become very limited in comparison. I still use DroneDeploy for processing regardless of what I fly with.

I don’t see anything described here that requires centimeter accuracy, especially when you are talking about tracking with a mobile device. Unless you are staking rows for plantings there has to be some other factor for that accuracy to be required. I would be willing to bet 50cm would be accurate enough and that can be accomplished several ways.

I would forget RTK on the drone and use GCP’s. You can use your existing M2 as a rover with NTRIP.

NTRIP is as accurate as a local base until it gets too far away.

Ok, I will reconsider Drone Deploy and take a look. I installed the hot shoe adapter on the camera flash which connects to the M2 for instant location stamping. i had strapped my cell phone to the aircraft and blue toothed to the camera for image gps stamps and set the camera to two second interval shots.
Our developers are requiring <5cm and I’m not gonna suggest other, as it’s their 7 year old baby and I"m not about to rename it for them. So they will do individual plant counts, height, volume, density, etc. But our main focus is locating a single bad anomaly out of an entire field and be able to pin point it.
We are not uploading to any cloud whatsoever, we are supporting a team of software engineers with uploads to their own portal so processing is out of the question. Until launch date, all real specific details are NDA protected, but some may already assume what our goal is.

Should work as well as the native non-RTK/PPK GPS on any drone. They’re basically the same chips. This setup with GCP’s will get you 5cm accuracy.

This is where you can only get so much help unless they agree to pull an industry professional in for consulting. Everything that I have suggested is centered around standard mapping that can also be used for Ag analysis as it is commonly used today. If they are doing something different than the 3 industry leaders then they have some R&D to do where noone will be able to answer the questions. When I started our drone program there was absolutely noone doing what I knew the systems were capable of so I know how intense R&D can become, almost to the point of whether or not it’s worth it…

1 Like

All great info and i truly appreciate the time you’re taking to help.
Our software team and company are in Europe and have mastered the RTK portion. They are already providing successful missions there.
It’s my job to deliver data from the US to process . So the learning curve is all mine.
My hurtles are: Learn how to deliver reliable RTK data that correlate to images.
Use a reliable flight planning software only to plan accurate missions and control aircraft effectively.

1 Like

Hi Dave,

I just want to add that you can take a look at our new PPK mapping guide as it might be of help to you.

This topic was automatically closed 100 days after the last reply. New replies are no longer allowed.