Originally Posted by kikislater

Good luck with your kickstarter campaign.
Be aware that emlid reach is already here and provide the same chipset m8t + extint function to trigger waypoint.
I think 3900$ is far away too much ....

I'm sure someone will correct me, but isn't the Emlid Reach RTK and the OBS just two different parts of the bigger picture that complement eachother?

The new RTKs for UAVs and rovers will provide a nice and accurate relative position, but they need GCPs to be global.
OBS provide GCPs at a much lower cost than traditional survey equipment?

Good day! Thanks for the question.

Ok so Emlid Reach is a component that you could use to build an OBS system (a SINGLE FREQUENCY version that is).

You still need to collect the data, process the data, power the unit, encase the units, buy survey grade antennas and possibly tune them to the enclosure, add batteries, injection mold, carry case, make and app to control them (if a person was trying to replicate OBS that is) then a PC program to post process those results using say RTK lib.

Emlid and Reach are just one component in a surveying system. That is what most people fail to understand... just because you buy a couple receiver boards does not mean they will be surveying anytime soon. Actually most likely anyone buying them on their own is to do so as a hobby or challenge.

Now as for using RTK on a drone this is different ball game. Many companies are out there claiming accuracies that are simply not true with RTK drones.

We have tested every RTK unit we could get our hands on and the only one with survey grade accuracy we have seen is the Mavinci.

Also just because Pix 4D tells you the data is accurate this is not always the case (usually not the case at all actually). You should do your own independent check using survey equipment (such as OBS) to compare a "Check point" on the ground to your created mesh. That is where you will see just how far off or close you are.

The reports generated by Agisoft seem to be more trustworthy but even then we sill do independent check shots.

So yes even though you may own an RTK drone you should still place a few check shots with OBS to prove your accuracy. And if you are on relatively small site (less than a quarter section) then heck why not skip the 40,000$ drone and use ground control with OBS

Does that answer your question? If not let me know.

Steve

Originally Posted by PaperCut

I'm sure someone will correct me, but isn't the Emlid Reach RTK and the OBS just two different parts of the bigger picture that complement eachother?

The new RTKs for UAVs and rovers will provide a nice and accurate relative position, but they need GCPs to be global.
OBS provide GCPs at a much lower cost than traditional survey equipment?

Lets also remember that Reach is single frequency. This is a completely different product than a dual frequency receiver.

Dual frequency has a faster fix time, faster convergence (this is the time it takes to accurately locate itself) and this is why we use them in OBS.

Single frequency is the dinosaur of surveying even though yes single frequency is subtantially cheaper it's just not practial for any type of work situation.

Steve

OBS isn't airborne at the moment? I thought OBS is a new way of producing GCPs cheaper and faster - yes? If that's the case, we still need a component in the sky that utlize GCPs?

Hi, no there is no airborne component needed.

You just use your phantom or inspire etc to fly the grid, then locate points on the ground with OBS and those points are used in agisoft, pix 4d, maps made easy etc to make the map accurate and also to prove your maps accuracy.

This will work with any drone data.

That's interesting . Instead of being dependent on the geotagging of the photos - just use GCP instead, and with OBS - plenty of those can be made in a sensible time? Is that the idea?
I'm sure you've mentioned in this thread previously, but what is the recommended max distance between the GCPs - given it's a relatively flat field (if the topology is relevant at all)?

Originally Posted by saabguyspg

(znipped some stuff)
...
So yes even though you may own an RTK drone you should still place a few check shots with OBS to prove your accuracy. And if you are on relatively small site (less than a quarter section) then heck why not skip the 40,000$ drone and use ground control with OBS

Does that answer your question? If not let me know.

Steve

Just on the 40k dollar drone topic, Tiny RTK for the new pixhawk 2.1 from Drotek is 600 dollar for the kit; base and rover (drone edition coming). Tuffwing has another - and that's the one I believe kikislater was referring to, even though he didn't take into consideration that dual frequency is a different world to single.

I've been mapping with pixhawk for a while, but only with DIY/budget solutions. As I was manually geotagging the images, I realised how inaccurate this process is.
Maybe more expensive solutions do it better - or... maybe processing services like DD, pix4d etc are not that dependent on geotags after all? That geotags are mainly a measure among several to make sure the images are placed correctly in the map?

I was thinking tiny RTK and OBS would be the perfect match, with good accuracy in all parts of the job.. but now I'm just curious if equipping the drone with RTK is bordering to be useless anyway?

I know my first response in this thread was "wish it was less expensive", but I'm slowly realising what a revolution to surveying you are about to introduce.

Yes exactly you don't need to geo tag your photos at all. However it's much faster for the software to align the initial data set if the photos have geo tags. Not to mention that it will make finding your GCP in the software much easier if the photos are geo tagged.

You mentioned "Plenty of those cab bee made in a sensible time?" Do you mean X's on the ground that are located with OBS? If so then YES! That's why we went with dual frequency. It will take you 3 minutes at each X instead of 15 minutes with a single frequency receiver that is 1/5 the cost. That adds up to a massive amount of time in one day

Steve

Originally Posted by PaperCut

That's interesting . Instead of being dependent on the geotagging of the photos - just use GCP instead, and with OBS - plenty of those can be made in a sensible time? Is that the idea?
I'm sure you've mentioned in this thread previously, but what is the recommended max distance between the GCPs - given it's a relatively flat field (if the topology is relevant at all)?

If a person wants to claim accuracy to a client and be able to stand behind it there is no way on board single frequency 5-10hz is going to cut it onboard a drone. You will lose fix and would have to fly at 4m / second (waaay too slow)

Not to mention if you want survey grade accuracy you will need to interpolate between epocs, shutter timing, lever arm correction and more.... so yeah... I would never try and sell any of those systems as "survey grade data" but it would be more like "better mapping data"

However yes add in GCP then all of a sudden you can prove your accuracy with the GCP and check shots but you are EXACTLY right... why bother with RTK if you are going to use GCP in the first place.

If you want to lose GCP and be confident in your numbers you need dual frequency quality receivers at a high refresh rate and calculations to locate your camera in the air to a few cm. After that you may still be asked to get check shots with your survey gear or OBS

There is a common theme her... you always need check shots. If you client is not asking for them now then you are lucky... as soon as you can't provide them they will move on to someone who can by using OBS or some other form of GPS equipment.

Steve

Originally Posted by PaperCut

Just on the 40k dollar drone topic, Tiny RTK for the new pixhawk 2.1 from Drotek is 600 dollar for the kit; base and rover (drone edition coming). Tuffwing has another - and that's the one I believe kikislater was referring to, even though he didn't take into consideration that dual frequency is a different world to single.

I've been mapping with pixhawk for a while, but only with DIY/budget solutions. As I was manually geotagging the images, I realised how inaccurate this process is.
Maybe more expensive solutions do it better - or... maybe processing services like DD, pix4d etc are not that dependent on geotags after all? That geotags are mainly a measure among several to make sure the images are placed correctly in the map?

I was thinking tiny RTK and OBS would be the perfect match, with good accuracy in all parts of the job.. but now I'm just curious if equipping the drone with RTK is bordering to be useless anyway?

I know my first response in this thread was "wish it was less expensive", but I'm slowly realising what a revolution to surveying you are about to introduce.

As for accuracy and spacing well good question.

It will depend on the desired accuracy in the end but if you are 100m between GCP you are usually good. Minimum of 5 required with two check shots. It only goes up from there.

Can you imagine carrying around 30 smart targets? No thanks. Especially considering you are usually surveying land that is not flat.... the reason we invented OBS was to carry less stuff to survey not more.

Now don't forget to add one in a the high point and one in a the low point of the site.

GCP should be spread evenly over a site. 5 in one spot is useless. The farther you get from your gcp in your produced model the worse the accuracy is in the data.

We are looking at developing some really cool tools for GCP placment that I can't talk about until we get funding.

The challege is that this is an expensive kickstarter. Most people want to support $900 kick starters not $3900 USD ones..

We have a lot of people who want OBS NOW but currently with Stryder development we can't afford the big chunk of cash to do the injection mold and clean up the front end (app) of the software.

So spread the word we are the right people to do this, we have more than enough experience in the real world of drone surveying to make this a very useful product you will love.


Steve

Originally Posted by saabguyspg

If a person wants to claim accuracy to a client and be able to stand behind it there is no way on board single frequency 5-10hz is going to cut it onboard a drone. You will lose fix and would have to fly at 4m / second (waaay too slow)

A way too slow but if you watch ebee rtk or applanix in uav, it's not more than 5-10hz and it works. I assume they made interpolation between points ...

What is good with onboard ppk system is sparse and dense cloud are far away homogeneous than without

Ebee RTk we have proven not as accurate as mavinci or stryder. Mavinci uses 100hz

Ebee is dual frequency but I have to double check hz on their latest version. I think it's 20 and hence it's not as accurate. I do know that hey have put a lot of time and money into the interpolation and that is not cheap either.

The question is I suppose how accurate is "accurate enough" we will also conisder 20 hz dual frequency for Stryder when we do more testing but for now the high end Stryder uses 50hz.

We will never use single frequency because it's just not reliable enough for a pro system. Great for fun or mapping but not surveying.

I will report back about Ebee hz as we have all that info somewhere.

Kickstarter is live...

https://www.kickstarter.com/projects...ef=creator_nav

Hope to see you all there and surveying with OBS early next year.

Steve

So cool to see it out there Steve. I'm in, but will have to wait till the new year with my pledge. Will spread the word as much as I can!

Thank you, and thanks for the support both on Kickstarter, email, and here.

Steve