Our company currently uses drones for video and pics for job progress or to gather site info for bids. We are exploring moving into using drones for topo work to track earthwork progress. Can anyone speak to your experience in regards to the accuracy, workflow, time invested vs benefits obtained?
Do it!
Took me a minute to get ownership to buy in but after the first flight with piles and having a point cloud for existing conditions I got the good ol’ " Why haven’t we been doing this".
Just like anything you need to know the limits but with good ground control, low flights and a good camera you can get consistent results under a .10 ft, even under .05ft, which far exceeds 2 ft contours on existing plans and field topo in general.
I can set control and fly a 20 acre site in 1 hr to 1.5. Processing can take a while so I recommend using something cloud base so you can upload and move on to the next thing. HCSS, Pix4D, Propeller and DroneDeploy all have solid solutions. If you go cloud based invest in a sloid internet speed to cut down on upload and download time.
3 Likes
Pat,
We currently have a 2018 model Phantom 4 and a 2018 model Mavic Pro 2. I see some services I have looked into recommend the Phantom 4 RTK. What are your thoughts on the proper drone needed to get the .10 ft to under .05 ft you mentioned?
1 Like
I have a Pro 2 and you can get those results with this but it is ore leg work. I think the Phantom 4 RTK would be a good fit. I still have fingers crossed for a Phantom 5 RTK but have not heard anything yet.
How do you mean “more leg work”?
Brian, These are the typical results I am getting with the Propeller system and Phantom 4 RTK. This was my flight from 2 days ago
35+ acre project. 15 minute flight time using 2 AeroPoints (ground control) and 10 checkpoints shot with a GPS rover to proof the surface. A site of this size you will see the data processed in a few hours. They spec 24 hour turnaround on data but that is very conservative
1 Like
Ed/Pat,
How does your workflow look. Is this a single man operation? How often do you fly each job during major earthwork moving?
I’ve used DJI phantom RTK and Sensefly Ebee geo combo with fantastic results. As far as workflow goes, you will get 100 different answer’s with 100 different results that all most likely work to some extent. I deal with Caltrans in socal and still place GCP’s even with RTK/PPK. without GCP, they wont even talk to you about quantities. I typically fly the jobs monthly and they range from 5 Acres up to 2500 acres and try to stay at 2.5cm/pixels GSD.
I use PIX4D and PIX4DSurvey to process the data and store in PIX4D Cloud.
The link below is a pretty good read on accuracy assessment form Sensefly.
https://sensefly.com/app/uploads/2017/11/eBee_RTK_Accuracy_Assessment.pdf
Andy,
When you say Caltrans won’t talk you about quantities without GCP’s are you referring to using the drone to provide quantities for actual pay apps?
Brian,
Correct.
Unless its a final pay item, we have used drone data for payment. Rock exc is a big one and being able to visually show the rock and quantify it is most helpful. The issue is being able to send the data to Caltrans/owner and them being able to certify or recreate the calculation for verification. Since they believe photogrammetry is an art form, you’ll need the information to conform to their standards, which currently requires GCP. If you use the data for internal record keeping and tracking, you could, I suppose toss some Propeller targets down or RTK off a VRS (cors her in cali) or unavco. I always use GCP’s in combination with RTK just incase the data gets elevated to survey or engineering. We have to follow and AMG spec here as well, if the GCP’s are set to those requirements, anyone can grab my drone data, i.e. images, camera calibration and GCP Inor and recreate the the point cloud and develop their own DTM.
I understand. Thanks Andy.
How many GCP’s do you guys feel need to be placed per acre to give you the needed accuracy? We currently use the existing or original ground the engineer provides in a DWG file for our dirt calculations for the owner. If awarded the job we do our own topo after the site is cleared to confirm the engineer’s topo. Most of the time this is good but we tend to run into issues with LIDAR topos over wooded sites. Do you feel confident enough in the drone topo to use it to check against the engineer’s topo and or replace the engineer’s topo?
1 Like
What you described is our typical workflow as well. We use the engineered supplied data up to post-award. From that point, we do our own verifications. The amount of GCP’s depends on a few things.
A: What is the LOD needed?
B: how complex is the terrain?
C: How large is the site?
Like I mentioned previously, most of our work is with the DOT or some form thereof; I tend to adhere to their methods in the standards. If you want an excellent surface reconstruction that you can be confident with handing over, GCP placement must be pretty dense. I shoot for no more than 1000’ feet between GCP’s on a pretty flat site. 500’ x 500’ will give excellent results. If you go through the site and plan it out, paint targets that will last the project’s duration, you’ll be pretty locked in and can add points as needed. What I like about dense GCP placement is you will have the ability to process with some points and use others as checkpoints. Once you get a good flight completed, you can use that as a background DTM in your flight software to further help control resolutions, resulting in cleaner surface reproductions. We will verify the DSM in the field with GNSS rovers. Also, note that you cannot get better accuracy than the target’s accuracy. My opinions may be argued, but my experience is if you set GCP’s with GPS, your drone data will not be more accurate than the accuracy you placed them. With options like Propeller, you have to understand absolute accuracy vs. relative accuracy.
Below is the Caltrans photogrammetry information and a GSD calculator I use.
How much time and manpower are you guys using to get these results? How does this compare to Trimble’s IOS system that gets information from active machines on site?
Trimble Works OS system is good but there is a lot of noise , excavators you can get extra highs and lows when over digging or feeling a truck, that comes with this and you lack the color that you get with a drone point cloud.
Drone flight is a cleaner snapshot in time of the site.
Do you use drones and Works OS together to get the results you want?
Yes. The two compliment each other very well
We are currently exploring both as options. Of course all options are being weighed relative to benefit vs investment. What would you say the pros and cons are of each? If you had to do one or the other which gives the most benefit per dollar invested?
Good Morning Andy,
I came across this thread and saw that your flying on Caltrans sites. We’re having push back from our RE here in Nor Cal and can’t seem to give him what they want/require. Caltrans web site is a disaster and I keep running into dead ends. Any advice for making Caltrans happy?