Me too. I have seen this tens of thousand of dollar worth of equipment and have no idea what is it for. So seeing a thousand dollar phone potentially replace it is intriguing.. but in the end, I don't understand most of it 🤣
Haha, I invite you to check out my channel and learn more about Surveying. My objective with this channel is to educate tech savvy people about the fundamentals of measurements and encourage old school surveyors to embrace the new technology as a helpful tool for our industry!
Rami, I am a surveyor for my company and when i realized i could scan a good amount of are ON MY PHONE I went nuts. I mainly use it on our Retention ponds and SWPPP details in the field to better explain issues or show some pretty accurate data
One thing to keep in mind is that the farther you are away from the object being scanned, the greater the error will be. So if you try that same scan again, you may get different results. It’s been a drawback of Lidar as long as I’ve worked with it. Thanks for the work.
Difference in horizontal measurements can be accounted for by how vertical the survey prism is being held which would explain why the elevations were apparently more accurate. Small change in survey prism is going to have a much smaller effect on the elevation. All in all looks good for grabbing point clouds of relatively small and complicated features, say drainage headwalls etc but will struggle over larger areas due to drift, as suggested above. Great test though.
Sensor issues aside (this type of LIDAR, being infrared, notoriously struggles in sunlight) most of the error you are getting is down to the fact that the phone isn't giving you the raw data but an already processed output which is optimized for producing relatively small-scale watertight meshes that can be textured and used in AR &c. The real promise is if you could combine LIDAR and total station measurements _inside_ the SLAM computation; having even a very sparse control network to anchor the environment map over large distances could potentially eliminate drift entirely. In essence, you'd be combining the long-range accuracy of traditional instruments with very high density of datapoints from the LIDAR.
Reminds me of my Engineering courses. Great explanations and I still got lost. I’ll stick to “about 3 fingers” measurement systems though, always on hand
Very good video. As always the issue is that when it fails, does for a large margin. The pro tools have being tried and tested for ages, and pros know the envelope of functionality to a very good degree.
Hey Rami, former student of yours in the Surveying Intro lab. Can you make a video explaining what the ground/grid scale factor is and how it’s calculated? I’ve never really understood it well. Thanks and love your videos !
Good to see you again Ben. Grid/Ground has to do with the curvature of the earth. In a local coordinate system, (like the one I used in this video), you would apply grid coordinates, meaning the plane at which the points are collected is flat. Whereas larger projects that could be 1000 acres of land will probably use state plane coordinates and require the curvature of the earth be taken into account. This is when we use the Ground coordinates. I will probably make a Geodesy Course one day that will explain all that and the math behind it.
@@RamiTamimi looks like you need to do your research on this (: Grid = points on SPC. Ground = local coordinate system. Grid is NOT flat, ground is. Earth curvature is also defined by what elevation the project is on. Remember, earth curvature at 0.002 @ 300 feet.
@@liamcarper6517 That sounds correct Liam. Grid coordinates are models of Earth that account for the curved surface, there are many models to represent Earth more precisely at different locations around the world; ground coordinates are localized coordinates for relatively small projects, too small to care about curvature.
When there are so many manual measurements/inputs present, what you're comparing is not the two machines but rather your own ability to measure properly.
I always wondered what those surveyors were doing with those instruments growing up as a kid 40yrs ago, down the corner street and on roads. Now have a fair idea. Thanks for the enlightenment.
I never thought that LiDAR has such accuracy, I used a GY-VL53L0XV2 Laser Distance Module with arduino but IPHoN 12 's LiDAR IS So phenomenal. Just Subscribed and Waiting to see the same comparaison using iPhone 13 pro
I understood about 2% of this but wow…I want to know more now. So interesting. I just got done binge watching Black Mirror again too. This is how AI will be able to recreate such real worlds. So much data. It’s beautiful.
Few points/notes on making it a controlled test: - The points you're taken on both are governed mostly by human error, you should set up three fixed baselines (triangle) with three defined points you can select more confidently (retrotargets, nails, etc). This way can ensure you're selecting the same points from both sources. - The instruments you use are both governed by distances. Therefore use both instruments at an equal distance. - Likewise the staff >height + >level error => point error, so should shoot using reflectorless - Then you can compare baseline lengths without the need for placing in "the same local coordinate system" and their angular changes. - Do the test say 3 times. Then you can see how precise it is too at obtaining repeatable measurements. This seems great for small volume estimates though! I know the video is probably just light testing/interest and you probably know these things. But I work on really precise engineering projects (1mm accuracy required sometimes 0.1) and you have people telling you to use drones, scanners, and photogrammetry cause they saw some video or some salesman promises mm accuracy and without understanding processing really grinds you! PS. Using imperial on that tap killed me internally those increments were far too big!!!
This proves one thing for sure, with bit more improved post processing iPhones can achieve near perfect portrait shots and portrait videos too. The current cinematic mode is purely software but I’m sure third party apps will take advantage of LiDAR for better portrait video in the coming months
When I was a kid, I spent many cold days “on the dumb end of the tape”. My father still used log tables to do closures on private surveys. When I introduced him to a scientific calculator, I was immediately promoted! Survey Science in those days was not a university degree program. Technology has taken us forward from the days optical plumbs! Thanks for the trip Dow memory lane 🇨🇦
I always wanted to know how accurate the LiDAR sensor in the iPhones is. Know I’m not only blown away by the results but even more by your knowledge about this topic. Awesome video
Hi Rami, Thank you for producing content such as this! I find your videos informative, easy to understand, and straight to the point. Please keep these coming, especially more like this one about the capabilities and and practical uses for the iPhone LIDAR scanner. I'll be getting one soon to help produce a small scale site model for my back yard. I will import the point cloud into Vectorworks Landmark to create a site model to help me with re-landscaping design. Just for the record I am a retired Aerospace, Mechanical, and Civil (structural design and analysis) Engineer who is fascinated by 3d modeling of existing conditions for engineering, architecture, and landscape design/site planning. Your channel is among the best and most helpful I have found. Thanks for your hard work and great content!
Man John, you left me speechless. My dream is to deliver high quality content and connect with my viewers in an intellectual way. More videos coming soon!
You need to bring the Survey Total data into the point cloud... this would give a better idea of its accuracy. Then you could highlight what you picked v the Survey station... see if there is a point there... Also ft! really surveyors in the US don't even use Meters yet.
Hello Rami, I am a Geomatics Engineering student from Nepal. It's so exciting to see these kinds of contents over here. Got to know a lot of things... thanks a lot!
This is fascinating content. I know nothing about surveying but I've always wondered about the process. There aren't many accessible resources so I really appreciate you making this. Subbed.
Excellent video - thanks for sharing such an in-depth understanding and the result was cool, although it shouldn’t be surprising - both the total station and the iPhone, are using a very similar technology - lasers, receivers and a “time of flight” calculation. The total station of course, also employs ultra-accurate angle-measurement mechanisms, but the main distance measurement uses the same principal as the iPhone. The iPhone lasers are VCSEL based and therefore lower power, lower distance (light disperses in air). The total station can measure huge distances (hundreds of feet).
Disclosure: I do robotics and 3D localization using IMUs and rolling shutter cameras for self driving cars. The alignment method he is using is not fair. It aligns the scale of the two pointclouds. This will lead to articicially better results as the scale is quite difficult to get right. Also one should use way more points to align the point clouds in the least squares manner. You need to align x, y, z, roll, pitch, yaw (leaving out scale as that is interesting for us). Two points provide only three constraints. Some degree of alignment is provided by the iPhone as it estimates the true-up direction using its accelerometer. This means the alignment could be better than one would expect. But that is relying on a very different thing than LIDAR. Also, this is not a true test of the lidar accuracy as it also likely uses camera and IMU is a optimization-based tightly integrated algorithm. This tests the accuracy of the system a a whole. Not the LIDAR itself. But who cares I guess ;) Scanning a larger area would show much larger differences.
totally agree with you. IMUs are what the Iphone is missing. Drift is constantly happenning in a mobile application, and by showing this small scale, that 0.30 error will grow exponentially as the error drifts. so he is comparing apples to oranges, really
Main problem of those Lidar sensor are not direct accuracy but rather drift... When ppl think they're gonna have precise measurements of hallways, around houses, floorplans etc... Take a 20 meters walk and the drift will be insane.
Interesting you decided to use a 1.8m pogo rather than a 0.1m mini prism to measure the kerb. If I was looking to do baseline obs for comparisons I'd recon that being slightly off plumb at that height would have a much bigger impact than the accuracy of the instrument. Its only an 8' bubble on the pole and it's not stabilised.
Fascinating video, Rami! I have been wondering about the efficacy of iPhone’s Lidar system for use in surveying since purchasing one a few months back. Thank you for doing the experiment so we don’t have to! What app did you use for the scan of the curb?
Great video man. It’s about time someone did a real world test on how close LIDAR sensor on the iPhone is relative to actual high end or professional tools used as shown in your video.
as a 2 time drop out from the PhD of Geomatics Engineering, I knew some "smartphone" would replace entire department within a few years, thanks for proving me right.
What App did you use on your IPhone to scan the Curb and create the point cloud and export? I have IPad Pro that has same Lidar and interested in this procedure
“We’re going to compare the accuracy of a total station versus the LiDAR of an iPhone.” *whips out measuring tape to establish the height of the total station*
Thanks great video, it shows something I cannot test my self. The only thing I did not like is the distance at which the phone and the total station was measuring.
Thank you so much for doing this comparison!! I've been looking for one specifically done with a surveyor who can appreciate accurate measuring to do this test. Would you be able to do this test on a cross-section of a road (even residential like where you shot this video) where you include a longer section of kerb and include a road-side kerb inlet (stormwater drain, not sure what it's called in America)? Thanks so much for this video.
@@RamiTamimi Even after purchasing the APP that can convert the file to DXF for me I can still only capture about 60ish feet from the start.. not sure if the app has a limitation on how much data it stores. Cant be the phone as I have well over 100 GB of space although it is early with this type of 3D scan from a phone i don't think the developers imagined people would 3D scan long distances
I’ve always wanted to learn what and how the surveyors are doing with their tools out there, and finally thanks to your extensive video, I still am clueless even though I enjoyed your content!
Thank you for the video. I am in attorney and sometimes we need to use Surveys to help resolve disputes. Thank you for the information, I will keep an eye out for future programs!
Interesting video! Another thing to think about - is the Iphone's exported point cloud based off the phones lidar measurements or an abstraction based on the generated 3d mesh. If it is an abstraction based off the mesh - mesh quality and tessellation settings may add additional error.
Super excited about all the accuracy discussion, then he pulls out an imperial tape measure "5.24" , Would be great to have overlays on the video in metric!
The only problem in building roads, Airports, underground excavation , Airports had tolerance of 0.05 on runways and taxiways. Highway work was +- 0.10 ft. Subgrade. Finish work was approximately 0.50
Why don’t you use more than three control points so that the two different local system can be overlapped more accurately. For example the Three-Parameter Transformation Formula or Seven Parameter.
