Surveyor, educator and PhD student Rami Tamimi walks through five years of testing the iPhone's LiDAR sensor for surveying, from the iPhone 12 Pro to the iPhone 16 Pro, each time checked against total station and GNSS ground truth. The talk explains the sensor theory behind iPhone mapping, the difference between relative and absolute accuracy, and what extras such as RTK or a scooter-mounted phone change.
“If we can get the iPhone to be accurate enough and we're able to trust the data that comes out of it, why wouldn't we be able to do this?”
“I think iPhone mapping in general is a good supplement to other types of terrestrial scanning.”
“if you have a good sensor in your phone, you should try to use it for mapping and see how it goes.”
In Rami Tamimi's tests, relative accuracy was mostly within two to three centimeters across iPhone generations, while absolute accuracy without RTK was off by a meter or more.
Rami does not think so: iPhone mapping is a good supplement that fills gaps alongside terrestrial scanning, not a replacement for surveying.
According to Rami, an RTK receiver brought the absolute accuracy of a slope test to within a few centimeters, but the relative accuracy of the LiDAR data stayed about the same.
Transcribed automatically (ElevenLabs Scribe) and lightly edited. It may contain small errors. Click a time to play the video from there.
Rami Tamimi: How's everyone doing today? Woo! Woo! How's everyone doing today? Woo!
Rami Tamimi: Woo! We still didn't have lunch yet, so you can't be that quiet. For those of you who don't know who I am, my name is Rami Tamimi. I'm a professional surveyor, um, a geospatial researcher and educator, and a PhD student. I've been on this PhD journey for some time now, so hopefully we finish that up soon.
Rami Tamimi: Um, so my, my presentation is very similar to, um, what was mentioned, but actually I'm going to be exploring the mapping with iPhones and the multi-year accuracy assessment evolution that we have gone through as we've tested the iPhone for its mapping capabilities. Now, I know that's a very controversial topic, especially among surveyors. So to have a surveyor talk about it, I think is... Well, I think it's interesting, right? Earlier this morning, we had asked a question about, uh, what's your job title, and number one was CEO, but I'm pretty sure number two was surveyor.
Rami Tamimi: So by show of hands, how many of you guys here are surveyors? All right, so we've got a large number of surveyors. That's amazing. I also noticed that twenty-eight percent of people acknowledge that they have skills in surveying. I would beg to differ.
Rami Tamimi: I think most of you have more skills than you realize, and we're going to explore those skills by assessing accuracy of the iPhone throughout many years, and to show you that, hey, you know, we gotta start somewhere. It's not gonna be perfect the first time you check accuracy of a dataset. It's gonna evolve, and it's gonna get better, and you're gonna get new ideas. So these are some of the things that I do. Among them, I am the founder of The Survey School.
Rami Tamimi: Uh, it is an online education platform designed for workforce development and educating anyone that is trying to get into the field of surveying. We have over a hundred students. Uh, we also have, uh, a large number of students outside of the United States, here in Europe and in Asia. Um, and furthermore, uh, YouTube is my biggest platform. I have over a hundred and fifty videos, free education online about surveying, so you're welcome to check out those platforms, um, after the presentation.
Rami Tamimi: So the big question is, should surveyors be using iPhones? Is, is it really a professional tool? Should we be measuring property corners like this and assessing boundary lines and writing legal descriptions like this? Is this something we should be doing? Well, perhaps.
Rami Tamimi: The real question is, how accurate can we get it? If we can get the iPhone to be accurate enough and we're able to trust the data that comes out of it, why wouldn't we be able to do this? Sure, this is an AI-generated image, but maybe there's a future where this is actually a reality. You see, I picked up the iPhone in twenty twenty when Apple introduced the LiDAR sensor, right? This was the first year I started seeing some geospatial sensors on the iPhone, and I started to think to myself, "Why?
Rami Tamimi: Why are these sensors on here?" And more importantly, can I use these sensors for geospatial mapping and for surveying? I have been on this rabbit hole now for five years, trying every new iPhone that comes out to see if I can improve the workflow, if I can see some improvement in these sensors, if this consumer device can actually replace, can it supplement, or can it actually be an alternative to many of these sensors that we see today, sensors that can cost hundreds of thousands of dollars. An iPhone can cost about a thousand dollars. And as I was developing the workflows for all of these experiments, I realized that there are four main components.
Rami Tamimi: Component one is the system. That's the iPhone Pro that I'm using. Every year there's a new one. Sometimes I do multiple experiments a year, but it's the same system that I'm experimenting with. Next is the environment.
Rami Tamimi: This is where I'm at. Is it an urban area? Is it a forested area? Is it an open field? These are all things that play factors into the way the sensors perform, and if that sensor doesn't perform well, how do you expect your data to be accurate?
Rami Tamimi: Next is the ground truth. What sensors am I using to validate the accuracy of my iPhone? This is extremely important because if I don't have a sensor that's accurate, how am I going to test my iPhone's accuracy with an inaccurate ground truth dataset? And finally, reporting on the relative and absolute accuracy. What's the difference between the two?
Rami Tamimi: How are they used? How do we know if the accuracy of the relative model is good but if the absolute model isn't? We'll go over that. Okay. Let's talk about some sensor theory.
Rami Tamimi: I apologize, I'm an academic, so there's gonna be some math involved. Okay. Let's start with inertial measurement units. I promise you, I will not go crazy with the math here. But an inertial measurement unit, or an IMU, is used to calculate the orientation and position of your device.
Rami Tamimi: Now, by utilizing the rotation matrix, which is the change in position, you're able to then see at a various time where you are. Now, again, I won't break down this too deeply, but as time moves on and as we rotate our IMU, we're able to see the orientation from start to where we are. That's T, time. The positional estimation- Is how far did we move? Did I move this way?
Rami Tamimi: Did I move this way? Did I move fast, slow? Based off of this information, we're able to actually model where the IMU is. Next is georeferencing. Consider this your position of the phone translated based off of the scale factor, the rotation matrix, and the translation vector of your origin point, that's when you hit start to record your data, and projected into some coordinate system, whether that's UTM or if you're in the United States, that's NAD '83, whatever your client wants.
Rami Tamimi: It could be a local system. Georeferencing is critical for absolute accuracy. Next is the LiDAR range equation. This is the equation that we use to figure out the distance from your LiDAR sensor to your object. It's a very simple equation, and it uses the distance of the object, the speed of light, because you're using a light pulse to measure, and the time difference it takes for that pulse to return.
Rami Tamimi: Now, there's obviously more math involved if you have multi-return sensors, but for the sake of keeping this simple, that is how LiDAR works. And sensor fusion is the mechanism used to bring all of this together. By fusing all of these sensors together in real time, you are now collecting data that you can use to acquire information about your survey. Okay, I'm done with the math. Let's talk about how I did these experiments.
Rami Tamimi: Twenty twenty, picked up the iPhone 12 Pro, and this was a hot topic on LinkedIn. Everyone was talking about it. Everyone wanted to know, like, "There's a LiDAR sensor. Can we use it? Can we actually collect data with it?
Rami Tamimi: Is it like a scanner?" And it's only a thousand dollars. Like, it can't be that good, right? So that's when I decided to take to YouTube and said, "I'm gonna make a video explaining how I would test the accuracy with my total station." So I started the survey by establishing a unknown point by resectioning to another unknown point, and using those two distances, I established a local coordinate system.
Rami Tamimi: This local coordinate system merely is a line that goes straight north. So now I've got this relative l- local coordinate system set up. I'm going to measure something. I didn't know what to measure, so I found a curb next to my condo and decided I'm going to measure this. This is not my twin brother.
Rami Tamimi: This is just clever editing. After taking the ground truth model, I then scanned with my iPhone. I used the 3D scanner app. The Lan Brothers had developed this app for fun, mind you, for fun, and they have millions of users now. Um, they don't charge for it, and they can extract point clouds and all sorts of deliverables.
Rami Tamimi: So I took this point cloud that I just generated, and I brought it into AutoCAD Civil 3D. It's the software I'm used to using in surveying, and I decided I'm going to try to map out the feature. So I used the grading elevations to figure out where the low point and the high point of the curb were, and I simply just pulled the curb and gutter data. Couple of points, nothing crazy. I didn't really know what I was doing.
Rami Tamimi: But that was the first test I ever did of the iPhone, and that video got over seven hundred thousand views. So this broke me into the rabbit hole that I'm in right now, which is trying to figure out how accurate I can get this iPhone to be. Okay, next year, twenty twenty-one. New iPhone comes out, second generation. I'm excited.
Rami Tamimi: I buy it. I've never bought a new phone before. I used my wife's phone, actually, for the year before. But this year, I actually bought my own phone to, uh, to be able to make this video. And I decided I'm gonna do more than just a curb.
Rami Tamimi: I'm gonna do this little island in front of my parents' house. I got a comment that also said, "Why are you using a two-meter pole? You should be using a short pole." So naturally, I took the feedback and went and bought this small little pole and decided to try to use it, right? Okay, why not?
Rami Tamimi: So I measured a bunch of random features: edge of sidewalk, manholes, PK nails, whatever I could. Just put some random things in there, measure it, and let's see, uh, what we can get when we scan it with the iPhone. Again, used the 3D scanner app. Started to play with the manual settings a little bit. Tried to get, you know, some more accurate scanning done.
Rami Tamimi: Um, and I generated this little model. Brought it into Civil 3D, extracted the points. All right. I've got my model for the iPhone 13 Pro. I promise we'll go over results at the end.
Rami Tamimi: Like, don't worry. Um, there will be numbers. Okay, next year comes. iPhone 14 Pro. I'm super excited because now I'm working with Leica.
Rami Tamimi: Thank you, Leica, for sending me an MS60. It's a half-second ac- angular accurate total station. This thing's a beast. One of the comments I'd gotten was, "You should be really testing a longer distance, just a linear distance to see how much drift there is in the data." So I did just that.
Rami Tamimi: I measured a hundred feet. I placed a target on every ten-foot interval, and then I measured it with the total station to get my ground truth data. Additionally, I decided I wanna test the iPhone 13 Pro with the iPhone 14 Pro. So I scanned a little park, you know, playground thing just to see how the reconstruction is, how the accuracy is, you know, just to kind of do some more testing than just one little scan. What I found was the iPhone 14 Pro created more vertices than the iPhone 13 Pro.
Rami Tamimi: Different LiDAR sensor, sl- slightly more upgraded. I wasn't a sensor guy yet. I was just trying to figure out accuracy for the system. Okay, iPhone 15 Pro. I'm gonna go a little bit fast here 'cause I'm short on time.
Rami Tamimi: iPhone 15 Pro comes out. What are we gonna do? Well, we're gonna measure this island, and then I'm going to take a GNSS receiver and establish geodetic coordinates. Now I'm gonna ground truth my data to geodetic coordinates so that I can measure the absolute accuracy. I then brought my total station in, imported the geodetic coordinates, did a grid-to-ground conversion, and began acquiring with a TS16.
Rami Tamimi: Slightly less precise, but still very accurate total station. And the last iPhone that came out was an iPhone 16 Pro. I decided to survey an entire building. Went all the way around the building, measured the curbs, the gutters, the building corners, as much information as I could, and then I actually ran a topographic survey, a full topo survey, with my total station. We generated the point cloud on Pix4D Catch, and I was able to actually use their snapping feature, which adjusts your dataset right after you finish it to ensure that it closes on where you started by using some matching of the tie points.
Rami Tamimi: The Civil 3D model looked good. If you take a look at the contour lines, they were matching, so that told me that relatively speaking, my model looked good. And so what do the results look like? Well, let's go back to the iPhone 12 Pro. I used two of the points to align my dataset to the total station so that I had some commonality when I would compare points.
Rami Tamimi: So I omitted points one oh two and one oh seven. By omitting those points, I came up with between some areas were a centimeter. I had one spot that was twenty centimeters, but in my mind, I was like, "Ah. I mean, I'm getting centimeter-level accuracy. I'm, I'm pretty happy with this."
Rami Tamimi: The following year, I did the same process, right? I just selected a couple of points, rotated my point cloud to, to match the, uh, total station data, and then measured the relative accuracy. I noticed that the accuracy was a little better here. We reached seven centimeters in some spots, nine centimeters in one spot, uh, but the majority of these points were within two to three centimeters. For an iPhone, I felt that was pretty good.
Rami Tamimi: Okay, but now let's talk about a long linear path. iPhone F- 14 Pro in twenty twenty-two, started measuring these points, and I found, again, I'm within three to four centimeters. So it's very similar to what we had the year prior, but now I'm actually really putting it to the test. I'm really making it travel a ma- a longer distance. When I reconstructed the data of the playground, I measured the absolute accuracy, and I found that the iPhone 14 Pro had a r- absolute accuracy of one point four six meters, whereas the iPhone 13 Pro had three point seven five meters.
Rami Tamimi: I repeated this multiple times and had very similar results. So I found that the absolute accuracy was actually slightly better. If I were to guess, I would say the trajectory data was probably better because the GNSS was better. iPhone 15 Pro, we measured that island. You can see I have two surfaces on the bottom here, um, and they-- One is from the total station, one is from the iPhone.
Rami Tamimi: They matched a volumetric measurement of about ten percent. Okay, not terrible. It's good for voluming, for doing volume work. Uh, and then the linear accuracy was between two and three centimeters, so pretty consistent now from what we've gotten from years prior. We also compared the 14 and 15 Pro.
Rami Tamimi: We found that both devices were about five meters of range, so the range was about the same. Didn't really change between those two. However, the density of the iPhone 15 Pro was greater, and once you got to five meters, uh, you start to really see some, uh, the structure deform with the 14 Pro versus the 15 Pro. The iPhone 16 Pro, I did run a relative accuracy assessment on it, as well as absolute. Again, two to three centimeters in the, uh, relative and about, I think we said...
Rami Tamimi: Where was it? Oh, I don't have it written here. Or no, I do have it right here. Um, in some areas, it was ten to twenty centimeters, but the worst of it, which is from one extreme to the other, was a little over one point three meters. Again, this is an iPhone.
Rami Tamimi: This is a one thousand dollar device, and it's doing pretty good. We ran a couple of special experiments, just things to test out. The first was attaching RTK to the iPhone to see if we can improve the trajectory data in order to i-improve the overall accuracy. So we did, again, a sloped measurement. This was a deep decline in Ohio State's campus.
Rami Tamimi: Uh, we measured it with a total station. We scanned it, and we found the accuracies were... Uh, this is absolute accuracy, not relative accuracy. Within zero point three to three centimeters. So absolute accuracy is definitely more valuable if you're trying to georeference your dataset.
Rami Tamimi: However, the relative accuracy didn't really change much, which tells us the LiDAR sensor performed the same regardless of the trajectory data being higher accuracy. Few years later, Pix4D reaches out and says, "Hey, we have, uh-- We've opened up our network. We are using a lot of other GNSS receivers, and we want you to kind of play around with adding data to an existing point cloud." So I went and scanned a new construction on a site that had been developed, and we have this site right here. Now I'm able to add this data in pretty accurately.
Rami Tamimi: We don't have numbers here, but I just think visually it looks very nice. Another thing I was able to do is take a, a utility drawing, digitize it into Civil 3D, upload that to, um, Pix4D Cloud, and then use AR to actually see the utilities out in the field. So again, some special experiments we did, some fun stuff that we did. Um, I thought this was really interesting just because I do a lot of utility work myself. And the last fun project I did was I put, um, my iPhone on a s- electric scooter, and then I ran two ground truthing systems, one with a GNSS receiver and one with a prism on a total station.
Rami Tamimi: And the idea was that I would drive the scooter, um, one a-at ten kilometers, one at twenty kilometers, and one at thirty kilometers an hour and see how well I can collect data and test the trajectory estimation, uh, accuracy. So here's me, uh, on the top with the GNSS receiver as a ground truth and on the bottom with a total station as the ground truth. Again, these-- the GNSS receiver is a GS18 I. It has an IMU in it, and then the total station is just being tracked by the-- or the prism is being tracked by the total station. So the trajectory estimations, this is the GNSS receiver.
Rami Tamimi: It was pretty good actually at ten kilometers an hour. It only struggled when it was near the trees. Down at the bottom is twenty kilometers an hour. You can see there's more struggle, um, especially when we would go under the trees. You saw that environment.
Rami Tamimi: It's a pretty urban and mixed, uh, environment. Total station held on pretty strong. Only one area in the twenty-kilometer-an-hour section where it actually struggled. Thirty kilometers an hour is where we saw some major problems. This is the GNSS receiver's trajectory estimations.
Rami Tamimi: Again, it's all over the place. It's not doing that great compared to what the iPhone measured. And this is the total station. It actually m- disconnect-- not disconnected, but, like, lost lock for a majority of the, um, scooter ride. So I estimated that, you know, if you go between ten and twenty kilometers an hour, you should be able to, uh, map out pretty accurately.
Rami Tamimi: And then we did a point cloud analysis of the datasets that we collected. Again, between one and two, uh, c- uh, one-- uh, ten and twenty kilometers an hour, we were fine. At thirty kilometers an hour, we were just completely off. So what's next? Well, we anticipate the new iPhone, whatever they call it, probably 17, will have more integration with external IMU softwa-- or hardware.
Rami Tamimi: Uh, we want to start using terrestrial laser scanners. This was something I wanted to do for this presentation, but we didn't have the time to do it. But we anticipate to do, uh, more sensors using terrestrial scanners to do the ground truthing, um, and plethora of other things that we wanna do in order to improve these experiments through the years. So if you'd like to learn more about surveying, you can check out my YouTube channel. You can check out thesurveyschool.
Rami Tamimi: com. Uh, and I thank the organizers for having us here and joining us, um, and thank you for listening.
Host: Rami, great job. Great job.
Rami Tamimi: Yeah, appreciate it.
Host: Great job. So we've got, uh, actually, once again, a lot of questions coming in from Renata and others. So let's bring them up on screen, and let's start with the first one. I would, I would merge the first two, you know, merge two-- first two. So the, the first one, how close are we to using iPhones for serious site documentation?
Host: Are they good enough?
Rami Tamimi: So this is my opinion. I think iPhone mapping in general is a good supplement to other types of terrestrial scanning. Um, I don't see it taking over traditional scanning. I don't see it, you know, being a replacement for surveying. Uh, but I do see it as a supplement.
Rami Tamimi: I do see it-
Host: Mm-hmm
Rami Tamimi: as us being able to fill in the gaps. Um, this is something that's happening today with RTK technology, with, with higher-end IMU sensors. Um, it's a matter of just having them integrated into our phones now.
Host: Got it. Um, next one from Ivan. Uh, do you think that mobile or wearable devices will replace all surveying tools used for 3D scanning in the future, in the near future?
Rami Tamimi: I do not.
Host: You don't think so?
Rami Tamimi: No.
Host: So as you said, like complements.
Rami Tamimi: It will complement, yes.
Host: Cool. Let's take one more, uh, one from Nina. Have you also considered extending this experiment to other mobile phones, uh, with similar sensors-
Rami Tamimi: We should
Host: so not just iPhone?
Rami Tamimi: I, I would love to try other sensors, other phones. Um, there are many phones that are banned in the US that I can't try. Uh, but there has been a, you know, a huge jump in increased quality of sensors. They're getting a lot cheaper, so they're adding them into our smartphones, and quite frankly, if you have a good sensor in your phone, you should try to use it for mapping and see how it goes. That's part of the surveying process.
Host: Rami, thank you so much.
Rami Tamimi: Thank you. Appreciate it.
Host: Yeah, absolutely. Give him one more round-
Rami Tamimi: Thank you, guys
Host: of applause, ladies and gentlemen.

Researcher & Educator
Rami Tamimi focuses on integrating field surveying with advanced technologies like LiDAR and photogrammetry. With over a decade of experience in surveying, spanning land development and autonomous automotive technology. He leads the largest online surveying community on YouTube and founded The…