My Day with the I-Site 8800 LiDAR Scanner

WOW!  It’s been almost 2 months since I’ve made a post.  Darn work getting in the way!  At least the work involved playing with some mining tech toys though!

Back in Mid-November, I had an opportunity to use an I-Site model 8800 LiDAR scanner, manufactured by KRJA Systems/Maptek, to perform a contract survey job for a mine.  I have certainly heard good things about the 8800 scanner, having seen it at its first unveiling at the SME show in Tucson in 2010.  I was never particularly impressed with Maptek’s initiall offering of the 4400 scanner, as it didn’t have much range and the data return was quite noisy, particularly on light-colored surfaces.  Given that preconception coupled with my largely positive experiences with Riegl and Optech scanners, the 8800 had some rather large hurdles to overcome.  After a day (really, more like an hour) of use of the 8800, my experience with the scanner can be summed up in one word – AMAZING.  It is absolutely the real deal – fantastically easy to set up and run, the data is clean and precise, and it is FAST!  I say I used it for about an hour, but was able to get 4 scans from 3 scan areas done, more than enough data to cover my area of iterest (at a range of up to 1500m!)  I’ll tackle the individual components of the scanner experience one-by-one, with some pros and cons included.

(Unfortunately, I’ve never gotten permission to use screenshots of the project for this blog, so for the moment I’ll just explain the functionality of the unit, and hopefully add to this post later if I get permission to use images.)

Set-Up and Operation

Choosing a Scan Location and Setting it Up

Most scanners I’ve used in the past are a bit cumbersome to transport, particularly over rough terrain.  While I won’t say the 8800 was a breeze to carry, a particularly nice feature of it is the fact that it doesn’t require multiple boxes – everything, including batteries, is inside the large Pelican case it ships inside.  This is a big difference from the 50 lb. deep cycle marine batteries I used to have to carry over muck to use the old Riegl pan-and-tilt, or the Optech scanners that can’t run on anything other than a constant current from a vehicle.  (I understand some of the Riegl units now have integrated batteries which is certainly a step in the right direction.)  Once at the setup site, it’s a simple matter of setting up a tripod as normal, getting the tribrach “roughly” level, (apparently the 8800 has an automatic leveler that will work up to 18 degrees, so roughly level is even extraneous,) and setting the scanner on it.  Put in the mil-spec battery, plug in the ethernet cord to connect to the tablet PC, and turn it on.  The boot time for the 8800 is a matter of seconds, compared to the couple of minutes the Maptek 4400 would take.

Maptek 8800 Scanner. Image taken from i-Site website - http://www.maptek.com/products/i-site/i-site_8800.html

The tablet PC is a ruggedized model (Panasonic I think, but I don’t recall the brand) that runs Windows with a custom interface that boots on startup, so it immediately opens into the I-Site scanner control panel.  In fact, I’m not sure if there’s an easy way to get it to boot into normal Windows, which is exactly the idea – it’s intended to only run the scanner, which it does quite well.  The interface is extremely intuitive, and there were no issues with connecting it to the scanner or to a USB drive later to pull the scans off (more on that later.)  I don’t have a screen shot of the controller, but there is one on the official informational brochure for the 8800 scanner which can be found here.

Backsighting

So, literally 3 minutes after selecting a scan location I was ready to run a scan.  My first scan position had been surveyed beforehand, and a single backsight point was set up roughly 180 degrees from the area I was interested in scanning.  To backsight the 8800 is unbelievably easy – simply point the scanner at the backsight, get it in the crosshairs, and tell the unit (via the tablet) the name and coordinates of the current location and the backsight location.  That’s it!  I actually messed up the first time I tried it, because I was expecting something more complicated.  The first backsight was also all that was required (even though I continued to locate the backsight in the subsequent in-fill scans, which I think helped the point merging but can’t say for sure) as post-processing will use coincident points to merge the subsequent scans with the first, geolocating all of them quickly and accurately.

Operation

Maptek has stuck with their workflow (pioneered in the 4400 and similar to Optech’s approach) of taking a photo image of the entire 360 field of view, then selecting the scan window from that photo reference.  Personally, I prefer the way the Riegl pan-and-tilt did it, where I physically moved the scanner to the “bottom-left” and “top-right” positions of the scan window.  I find the “fisheye lens” effect of the scan window to be difficult to discern the area of interest at times.  However, I would chalk that up as one of the few “Cons” of the experience.

One of the huge differences between the 8800 interface and those of other scanners is the simplicity – on the Riegl scanners I’ve seen, you set things like mgon differentials, angle separations, etc. to determine the point cloud density.  On Optech scanners, it’s necessary to tell the scanner the Range of the data of interest (which IMO is fairly absurd – isn’t the unit supposed to tell me the range of objects, not the other way around?)  None of this is necessary with the 8800.  I don’t know exactly what the “closest” range is, but it’s close (within a few feet.)  Likewise, it will shoot out to an impressive 1800m without batting an eye, and without having to change a single setting from the close to far points.  HUGELY useful.  Likewise, not having to guess at the angle separation I want for the point density is extremely useful as well.  I would imagine researchers and educational institutions might  like more control over such parameters, but for the everyday user (i.e. field surveyor) they’re at best excessive and at worst confusing.  There are 5 density settings that are more than sufficient to get pretty much anything you’d want (for my work, I only used the lowest and second-lowest point densities.)   The scan time f0r a 360 degree low-density scan was 3.5 minutes; the longest scan I took was just shy of 180 degrees on the second-lowest density and it took a whopping 9 minutes.

As I mentioned before – the entire time I used the 8800 was around an hour, INCLUDING driving time to the other 2 setup locations, because of it’s amazing speed.  In contrast, the pan-and-tilt Riegl would have taken about the same amount of time to do a 40-50 degree window with a fraction of the data points, and the Optech would have still been taking the photo of the first scan location (I kid, but not really…)

Data Handling

Obviously, handling I-Site 8800 data is going to be a breeze provided you’re using I-Site Studio software.  The first requirement is to get the scan data off of the tablet PC and onto a standard computer.  This is done using a USB thumb drive.  When the thumb drive is plugged in, the I-Site software automatically detects it and asks if you’d like to transfer the new scans to the external drive.  Simply click “Yes”, and it sends everything over in .3dp format to be loaded into I-Site Studio.  I think this is important, as I’m not aware of the other scanner companies doing anything quite so seamless.  Riegl has RiProfile, which is fine for obtaining scan data but is absolutely terrible for post-processing – horribly cumbersome and non-intuitive.  It’s a bit like using an etch-a-sketch to do CAD work.  Likewise, to my knowledge Optech doesn’t have any post-processing capabilites – they push PolyWorks software which is quite powerful but (at least when I used it many years ago) was rather difficult to use.  Also, because it’s a third-party connection, there isn’t the seamless integration a user gets with the I-Site system.

Once the data was transferred to a computer running Studio, it was extremely easy to get everything oriented correctly in space.  The first scan situated itself automatically, as it was backsighted in the field and that data carried over with the transfer.  Awesome!  The rest of the scans oriented themselves quickly and without fuss using the “global registration” option in Studio.  This option actually didn’t work for me initially when I first imported the scans.  I had to do a little bit of manual translation and rotation of the non-georeferenced scans to get them into their general locations.  Once I did that, the global registration worked really well –  it oriented everything relative to each other and to the georeferenced scan pretty much automatically.

A couple of additional hours were required to clean up errant points and add points inside shadow areas (the project was remarkably limited in access areas to get decent downward scans in a lot of areas, so benches and roads ended up shadowed where berms etc. were in the way.  Additional points helped fill in the roads and benches at their proper elevations, but were used sparingly enough that they shouldn’t have affected much of the area of interest.)

Speaking of data cleanup, the data itself was phenomenally clean – hardly any blowout points or random clouds off light surfaces that plagued the 4400 scanner.  In fact, the only real issue I found within the area of interest was that the registration of the scans left a slight gap from one scan to the next, resulting in “needling” of the final surface.  This was easily fixed using the Topography filter.   Otherwise, I’ve never seen data so clean and accurate, even from an Optech scanner.  I would be interested to see the two of them go head-t0-head someday…

Conclusion

I think it should be fairly clear that I am a fan of the 8800.  I would (and absolutely intend to) use one again for any job that may come my way.  Here is a synopsis of the pros and cons as I see them:

Pros:

  • Blazingly fast – from setup to scan time, this scanner beats all others I’ve seen by a mile.  Granted, I’ve not seen some of the newer offerings by other companies, but given the range and accuracy, this one screams.
  • Simple to Use – No one wants to be in the field troubleshooting a scanner issue in  the middle of winter, whether it’s hardware or software related.  The 8800 solves both problems with simple interfaces, intuitive controls, and solid hardware.
  • Post-Processing is Easy – integration with the established software package I-Site Studio is going to be a winner no matter what.
  • Photo Drape – Even as far back as the original 4400, I-Site has the photo drape figured out.  It’s great to see this continue in the 8800, and makes a lot of tasks much easier (such as geotech analysis.)
  • Truck Mount capability – while I didn’t get to test this feature of the scanner, it is apparently the new greatness.  They have a truck-mount assembly that attaches the 8800 to the roof of a truck with a couple of GPS antennas.  The GPS will locate and backsight the unit automatically, and the scanner does the rest.  Meanwhile, the surveyor simply drives the truck around, taking periodic scans, and at the end all of the scans are georeferenced and oriented together, doing the scanning and post-processing at the same time!

Cons:

  • Fish-eye image preview – as I mentioned before, it’s often hard to differentiate what to scan from the skewed image.  It would be nice if they could either un-fisheye it or give a more physical scan window alternative.
  • Photo Exposure Controls – though I love the photo drape, I think it’s high time to integrate some auto-exposure into the camera.  The scans I took were of snowy surfaces in bright light, and even on the brightest exposure setting they came back washed out and in some places unuseable.  If my smartphone can adjust itself for exposure, so should an expensive piece of equipment.
  • And speaking of expensive… – The 8800 is no slouch in this department.  At a price tag of roughly US$250,000, it’s not something an average Joe Surveyor is going to be picking up at Wal-Mart.  I’m not sure if it’s the highest-priced terrestrial unit of similar specs, but it must be close.  (Side note – apparently the newest addition to the I-Site arsenal, the 8400, is a much more modest $120,000.  It does not have photo overlay, but it has an integrated electronic compass so it can backsight itself.)  The 8800 units are also available for rental, so for periodic jobs that don’t require a quarter-million outlay, that may be your best bet.

For more information on the I-Site 8800 or any other I-Site scanner products, visit their website here.

 

About