Spatial Analytics

Members
  • Content Count

    14
  • Joined

  • Last visited

Community Reputation

1 Neutral

About Spatial Analytics

  • Rank
    Member

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Happy to report I have automated the marriage of DJI image embedded position data with the events in a post processed reach PPK file. This one step ahead of 'normal' photogramettric processing can reduce distortion and the number of control points needed for a high quality (yes phantom 4 and high quality can be in the same sentence) orthomosaic.
  2. Don't know about dropping like a stone, but as the drone heats up the altitude drops by 10 feet while reporting it is at the same altitude! I discovered this by mounting an independent RTK GPS on my P4 AltitudeGraph This is corroborated by looking at the flight log recorded by MapPilot, which always ends while the aircraft is at a REPORTED altitude about 3m ABOVE takeoff. The log stops when the propellors stop turning and the drone is safely on the ground...
  3. If you change the image size without modifying the camera parameters I would agree with you, another option is to vary the x and Y pixel pitch, depends what the processing software can handle. My reults have been adequate, but not if the drone is crabbing at all. It does sound like you have a far more comprehensive approach, visited the web page looks impressive. So yes I will dig out a dataset or two. Depending on mission I carry an Emlid Reach kinematic GPS on my Phantom 4 (with rolling shutter) and have developed a method of linking the photo timings to the PPK data. I have one recent mission where I also stopped at each photo location. I am fortunate to have a tightly controlled archery range as my proving ground, I can pick and choose reference points or GCP's.
  4. Agree with you entirely. As a drone operating company you make your money based on flight efficiency and once you go down that path it makes sense to deliver noncontroversial accuracy. For example, as a vineyard with rows spaced 6ft apart and individual vines 4ft apart along the rows pre-processing for PPK accuracy before running through automated mapping without the added layer of GCP's does give me the ability to look at a vineyard at a plant level. At 50 Ha (123 acres) per battery the P4 solution suits many small / medium owner operated vineyards. So as I said, I think we are in agreement, I am just looking at a different market that looks at accuracy in terms of their needs, not that of legislatively mandated numbers. Of course for any vineyard I would also advocate the installation or permanent machine readable GCP (as per ,for example, Agisoft Photoscan) at about a 40 - 60cm diameter depending on need. These would not necessarily be used every time.
  5. Here in Australia we have just had a change to the commercial drone laws. Sub 2Kg (4.4lb) drones need company registration, but no organisational or personal pilot training. This means the sub 2KG class will bloom and many tasks which were in the realm of the registered, certified and trained drone operating company will be undertaken in house by the engineers, farmers, realtors. I am working with a number of these to develop solutions that fit their positioning requirements. In these cases rather than mandate over precise (and therefore expensive) positioning requirements I work backwards from their end product. In many cases 30cm really delivers and allows a business to afford to fly daily or weekly missions to monitor change.
  6. Thanks, Agree with the switch to LIDAR , for my own investment in this space I have an order in for the Quanergy S3. Unfortunately my work is in Australia where the trees are eucalypts and pines, they have no winter periods. I am working on algorithms to select imagery for mesh generation only from the ground and flying very dense non vertical camera missions. This is specifically to build up imagery of the sub canopy for archaeology where the trees cannot be disturbed, but a whole site picture is needed.
  7. At high speed the UAV is likely to be pointing pretty much along the line of flight and the rolling shutter distortion results in the bottom of the image being collected 30ms after the top. My simple algorithm is to resize (reduce) the image vertically by the number of pixels represented by (distance travelled in 30ms)/(Pixel size at flight altitude above ground) example 10 metres per second = 30cm 60 metres altitude = Pixel size of 2.6cm 30/2.6 = 11.5 (11) pixel reduction in height. The timing of the photo is moved forward by 15ms (half the rolling shutter delay) Is your software doing anything more complicated? Of course if you are stationary the rolling shutter is not needed to be corrected for.
  8. MapsMadeEasy's ap MapPilot does a very nice Linear mission, if you use a Phantom.....
  9. With the price differential it is almost worth just buying a fleet of P4s and fitting them out differently permanently. Refurbed P4's are less than $1000. Adding an Emlid Reach post processed kinematic GPS to a P4 will set you back $400-500 and give you the kinematic accuracy, the unit is easily transferable between drones.
  10. The biggest grumble I hear about surveyors is their perceived (by their customers) reticence to adopt new technology and deliver the same old same old, justifying this by legislative acts. I am a surveyor, an hydrographic surveyor, we have had to adopt technology for remote sensing the bottom of the ocean and remotely / autonomously piloted underwater vehicles to meet our clients needs. Anyway credentials established. The DJI consumer drones (Phantom 4 series) represent a remarkably capable and highly integrated platform for aerial mapping at high resolution. I think we can all agree with that. At 60m having a 2cm pixel size is awesome. L1 RTK GNSS systems such as the Emlid Reach have massively reduced the cost of GCP setting, but still need care and knowledge to do it right (auditably and verifiably), but you can do it without GCP's if you mount the Emlid Reach unit on the drone and post process the kinematic position. This reduces position uncertainty to well below the metre mark. In fact surveys with the Reach on the drone have shown up an interesting feature of the DJI altimeter, as the platform warms up the drone drops two to three metres whilst reporting that it is maintaining altitude. Something you would not know without a separate sensor! This explains those not quite right sloping DEMS that people produce, but also points to a way to use a $1000 drone for 'professional' work.
  11. Until recently this was pretty good guidance. But then I was working in very hilly country with significant tree cover and deep canyons. Getting to spots for GCP setting was hard and the heavy tree cover made the time needed long and the results obtained poor. So I put an Emlid Reach unit on my Phantom 4. Took a while to develop the timing mechanism, but the results are pretty awesome, 30cm (1ft) RMSE. I post process the kinematic data then run a simple python script prior to processing the images in my normal workflow, except I don't need GCP's
  12. If you carry a raw GPS logger that is post processing kinematic (PPK) capable then you can get 30cm (1ft) accuracy from a phantom 4. This starts to blur those lines between classes and certainly gets around the Californian Professional Land Surveyors Act. And in California you have access to the RINEX files from a large array of base stations. I use an Emlid Reach GNSS unit on the back of my P4. It works really well and I have not noticed any degradation in flight characteristics, maybe a minute less in flight time. I am achieving 30cm(1ft) RMSE and my check points (not GCPS's as I don't tie my map to them, they are just for me as I verify the method) are all within that 30cm (1ft) radius.
  13. Now that it is possible to fly the P4 with RTK onboard the need for GCP's is reduced. I have just completed a mission where the RMSE is 0.35m from the photos with no GCPs and agreement with ground verification points is less than 30cm in all cases. NO GCP processing, just carry the sensor and run a few scripts prior to your normal processing workflow.