For ball park measurements calculating building materials required, this is accurate enough. For scanning between known fixed points, it is also good enough to generate a profile. Given that the maximum range is only 5 metres, you are pushing against the accuracy limits anyway. Pretty impressive performance from the iPhone given it is not mounted on a tripod with a fixed remote reference point.
We tried it and we're off by 8.6 inches over a paltry 61 feet. Worst part was that if we hadn't checked our work with a laser scanner, we wouldn't have known that we were that far off. If that's "in the ballpark", that's an awfully big ballpark.
@@chrislutes2882 as was said, it’s accuracy is 5m, not 61ft 😂 This is ideal for indoor mapping more than anything, scanning small rooms etc, not for large scale work, that’s not what it’s designed for.
@@RamiTamimi I drive lorries and I sometimes double check the height of a lorry before going under bridge or trough narrow gate and so far have not been dissapointed using measuring tool built in iphone 12 pro max
I’ve been making some crazy architectural diagrams with the LiDAR feature. Had it first on my iPad a year or 2 ago and blew some of my tutors minds back on my degree.
Me, too. Wish I had this thing 20 years ago when we were measuring a route for a new sewage pipe - 10km long, 2m wide, doing it with just a total station and no GPS. That LiDAR thing accuracy would have been way enough and time spent on measuring a fracture of what we took.
I once had to set a bunch of stakes on a property line for my girlfriend…. It was a pretty complicated property line, with odd arcs and angles and set on a steep hillside. I took a plat map of the property and used my iPhone for bearings and a laser ruler to measure starting from a boundary marker. I used a construction app to correct for elevation based upon measured angles to each stake as it went up the hill and calculating the base off the hypotenuse. And double checked with the GPS receiver on the phone for elevation changes. For arcs I set a stake at the center of radius and just measured off the stake to mark out the curves. The neighbor decided to hire a surveyor to check the property line and the surveyor found every one of my stakes to be within an inch of the true property line. He asked me how I did it and was astonished that I could figure it out with nothing but an iPhone and a laser ruler. iPhones are essentially tricorders.
@@M-PASTA That was 7 years ago. And I’m not with her anymore…. But even 7 years ago, you could use an iPhone to ascertain bearings and read GPS elevations. I used an app called Theodolite. If you have one known point in the form of a boundary marker, you just measure distances and bearings as listed on the map. The only hard part is correcting for elevation as the plat map is a flat projection. But the property map had topographic lines and numerous bench marks noted for verifying elevation. By averaging the GPS elevation measure with the nearest benchmarked elevation, I could derive a hypotenuse length based upon the plat map distance and measure up the hill that distance ( when you’re doing it on your own, you need a bright surface to target the laser ruler on and you need a laser ruler that can handle the 200 plus feet you’ll need to measure. So you plant a target at a known point and walk the bearing until you get the distance you’re looking for. ) You can double check your elevation estimate by measuring the angle to the prior known target with the iPhone’s inclinometer and deriving how much higher than the target you are with the same Pythagorean calculation. Then you go back down and fetch your target and relocate it to the new known point. My target was literally a length of galvanized electrical conduit with a piece of stiff white board on it with an X at exactly my eye height above a hole with a bolt thru it drilled perpendicular thru the pipe about 18” up the conduit and I would drive the thing into the ground with a short length of larger water pipe slid over the conduit until the bolt I was banging against was at grade- using the iPhone to make sure it was plumb. This way I am measuring from my eye height thru the iPhone to a target that is my eye height above grade. I literally used an improvised plumb bob to transfer the position of the iPhone to the ground. ( holding the upper end of the string in my left hand just below the lens of the iPhone ) As you compile points, If something is significantly off, you run a do over. Over plotting a whole bunch of points, you can tell if you’re on track by the cumulative error. It took a lot of traipsing up and down the hill, resetting stakes and target to correct for cumulative error… but overall errors over and under true measure tended to average out. For example, I would stake maybe 6 points along what the map showed as a straight line… and then sight back along them and correct to the mean line of sight that was on the correct bearing. I knew I was on track when my plotting placed me within a few inches of a corner marker hidden under a bush that I didn’t even know was there. All in all, it gave me profound respect for the guys who had surveyed the first ‘accurate’ height of Everest, by starting at sea level in India and working their way across hundreds of miles using nothing but a theodolite and slide rule. I would not use this technique to survey unknown ground, But with the backup of a plat and topographical map with known points you can cross reference, you can plot a property line reasonably well.
Just wondering, would having the phone attached to a gimbal like the Osmo help with accuracy? I can imagine a certain % of error could be due to shaky hands.
It might help a little bit but the pull rate of the sensors (gyro, compass, accelerometer) in the phone seem to be high enough to easily calculate out pretty much any shaking in processing. It’s really impressive for such a small device and I wouldn’t have thought that it would actually be that accurate before watching the video
Surely you need three alignment points for a 3D dataset, or your elevations will be off for all but the two alignment points and anything directly between them.
I am impressed with the accuracy of the iPhone-generated point cloud, especially considering the phone was not positioned or moved in a controlled manner. To take it to the next level, the LIDAR app could recognize objects placed on several KNOWN precise control points (as done in the video), the known exact locations of those (2 or more) control points would then be loaded into the app, and the app would apply corrections to all the data based on knowing several exact actual positions within the data. Probably could be worded better. End result would be quite accurate surface data using the iPhone sensor in conjunction with the total station or an RTK system.
Great Job Prof. Rami. I am doing some tests with my Iphone13 Pro and 3D scanner app. Only one question: how can I export the point clouds from the app in .rcs format? I can't find this function in the app. Thanks in advance
As a new real-estate agent in AZ, where so-called "mustang subdivisions" are scattered all over the place, having a strong understanding of surveying only helps our potential buyers to identify risks and possibilities BEFORE they sign a contract... Also, would love to see same comparison, but using the Matterport Pro2 camera system.
I greatly appreciate your videos, very informative and delivered in a way a layman can understand! I’ve been in the heavy equipment grading side of construction for over twenty years. The company I work for recently invested into gps technology, I’m the sole operator, I hope to train more, of the gps grader, rover, and I make our maps via Trimble business center. My previous experience with grading was laser with a grade rod. You have helped me tremendously with filling gaps of knowledge I was missing. It’s hard sometimes to ask a question when you don’t even know how ask it, much less even know your missing key knowledge lol. I look at brochures and promotional videos of the equipment we have to grow questions to look up. There’s not much info on TH-cam for this line of work. I really hope your channel gets traction and gains in popularity, I know it’s a big time suck to make these informative videos but please keep up the great work!
Thanks for the video. I'm a surveyor from Australia and was intrigued by the LIDAR on these phones. I'd be curious to see the results if you did scale the point cloud to the surveyed points. The errors look a similar order of magnitude the further from the base point. The error on the elevation though is a bit strange. Thanks for the metric much appreciated. If you want the table to look neater you can use whole numbers for mm. Only use decimals for super high accuracy surveys.
Jason I’ve found hard stand to be within 50mm and up to 200mm on “grassed” areas, there’s a bit more testing to do and a better methodology to follow but it’s getting there. I’m a surveyor in WA if you want to ask some questions or look at the point cloud
The elevation error could be caused by the AutoCAD align command. When 2 points are used for the ALIGN, the command asks for the scale option. You can also align using 3 points. With 2 points, the point cloud can be slightly skewed from its original horizontal plane, and it is aligned to a line rather than to the plane. A better way would be to average X,Y,Z coordinates for the control points, then do the same for the total station points. This would establish a center of gravity for both datasets. Then you could place them on top of each other using this centre of gravity. Similarly you could calculate average angles for the pairs of points to obtain an average azimuth for both data sets and rotate the point cloud accordingly. Then you could mitigate some of the total station error by taking measurements of all points from various foresight points. This would create an overconstrained network of points, which would allow you to average the errors using Truncated Taylor Series (long story how this is done, but it works magic). This way you would see each control point as 3-dimensional probability ellipsoid with a range of errors, as well as the center of this ellipsoid. Then you would be able to mathematically prove the accuracy of your control point and the point cloud error. Sorry for the long-winded and academic explanation of this problem. It would make a great project for an advances surveying class. Great video, very informative and the methods shown were sufficiently accurate! Thanks!
Hello. Work with measurement of indoor environments for woodworking. In your opinion, would the iphone have the necessary precision to measure a kitchen or a room?
I have an iPhone 12 pro. I would love to just know how to use my LiDAR scanner when I can do with it for every day uses. I tried watching this and I understood what you were doing with the 3-D app on the iPhone but other than that I didn’t understand anything. Especially when you get into the software I had no idea what you were doing. Maybe you can make a separate video and every day uses of the LiDAR scanner. Thank you I found you by your drone videos I am very interested in serving the drones. And my part 1 07.
It appears the local accuracy is better than you have concluded. The errors you are showing in Z are mostly from the registration process being a 2-point method. You will get much higher accuracies if you do a 3-point registration. i.e. the phone's "level" was off from the total station level, which is expected given it's coming from an accelerometer and not a 10" bubble. This put points that lie near the line extension of your two registration points in high accuracy (105, 106, 113, and 114), and points to the upper left in a negative Z error (101, 102), and the points to the lower right with a positive Z error (107-111). I took your points and did a 3-point registration without scaling, shifted the points for best fit by the average NEZ for all three control points, and then got the standard deviations: 0.10' E, 0.10' N, 0.04' Z, and max absolute errors of 0.19' E, 0.16' N, 0.06' Z. The remaining error was clearly a scaling error as the error directions all radiated from the center. The fact that the Z had less deviation is because the camera was facing down, perpendicular to the scan surface. If you want better scaling, I think some targets with vertical surfaces placed every 30' would give you that. Maybe try it with some lath stakes at the ends. Ping pong balls also make good photogrammetry targets, and would probably work with lidar as well. Another good test would be a route between two points and see what the real absolute error looks like. This test really only gets you local errors. That being said, it's looks like it would be a good tool for some applications.
Excellent demonstration. You really did put a lot into this video and it shows and for someone like me that hasn't Really seen a lot of this, it was a very good video and it taught me a lot thank you
Dang. Can you use the iphone lidar to create a 3D model of an object and then export to a filament 3D printer? Maybe even scale the object down to fit the print area?
I helped my dad do Survey work for our family's construction company back in the late 90s. How I wish we had a little robotic adjustment like that between the rod and station.
good video. Remember that good survey practice is to have your backsight as the furthest point away from your detail survey.... if possible to avoid angle error. Also you had high vertical angle readings, so RL error would be higher. Would be good to see your next video on this.
What errors if you transformed Lidar to 3 survey points? Short rod tough on the knees.. Needs an extension to standing height with the bubble at eyeball level, leaving the reflector low.
I think adding some more 3d features would help improve the height accuracy. Like by leaving some concrete blocks in a grid every 2m. The software likely struggles to tell if mostly flat ground is flat or slightly curving otherwise (when it's stitching together the various 5m sized frames from the video feed to model an area much larger than 5m).
Please, can you compare the iPhone LiDAR with the Terrestrial LiDAR Scanner, to determine if the iPhone will be appropriate for taken 3D structure or point of a plant. Thanks
I’m a real estate lawyer and actually had an appointed expert-surveyor in a case take a LiDAR scan with his iPad Pro, to scribble notes on and take rudimentary measurements from. And all the (many) parties asked about the app and how he did it. It was so funny.
Thank you for the video. I have all of the corner pins of my property marked. is a string accurate enough to use to set a fence on the 628'boundary line or would you use an instrument of some type?
Rami, this is the first video I’ve seen of yours and i found it to be fantastic. I’m not a surveyor, nor even in the construction industry. However, I do watch a lot of home building, construction, excavation, and concrete videos, so thats probably why your video came into my feed. This was so well explained I was able to follow and understand the whole process despite never having used either the survey tools nor an iPhone 13. Here are some thoughts I had. 1.) The Total Station obviously has a lot of advantages afforded to it, for example, it’s fixed position. The biggest advantage, however, I think would be the prism. 2.) While there will be movement induced inaccuracies from the LiDAR sensor and accelerometers in the iPhone, I think there may be resolution and backscatter inaccuracies in the point cloud. 3.) A set of inexpensive markers could be made with a bubble level and some object the LiDAR could accurately key off of (a small orange ping-pong ball, steel ball bearing, 6-sided die? obv. testing needed). 4.) With an effective marker in place, the survey points in the point cloud would jump out and we can subtract the marker’s height from the measured elevation. In theory, i think this would reduce the prism’s advantages, and some other’s suggestions like using a gimbal or a weighted steady-cam could help reduce motion induced error. Thanks for the great video! definitely sub’d. 👍🏻
looking to use the iPhone Lidar to capture deflections of a concrete ceiling. Would you be happy to share your excel spreadsheet? I'm hoping to get an accuracy of 2mm or better, do you think that is possible using the iPhone?
Very good analysis. It was helpful, since I have been thinking to buy leica BLK 360, I see for generating rough existing plans with +/- 100mm accuracy this is great. Thanks man.
Perfect intro for a layman. Good analysis between accuracy and precision. Impressed that the IPhone software was able to stitch the cloud together. Fine for estimations; not suitable for a design survey. Additional control might aid in accuracy. Keep the land surveyor over the IPhone. But I'm biased.
Wonder how would the error fare on large area. IMU sensor [which phone use to detect its movement] has inherit errors which would accumulate under prolong use.
Not to be a stickler, but seems like the deltas are somewhat correlated with the direction and distance. Deltas between iphone and TS on local north grid are fairly tight while west and south, deltas are building up. In you check, we can't see in the video if a HChech and VCheck was performed ... Transit and Face to CP. We should not see a correlation between these deltas correct?
For the uninitiated, 0.01’ is about 1/8”, so I gather that the iPhone has about a 1”-2” accuracy for this scan. Probably good enough for estimating take-offs for materials and construction boundaries, but we will probably have to wait for the iPhone 16 before we’re aligning machinery with it. 🤓
Your error could be from skew of the scan. This should be done with picking all the points in the scan coordinate system and labeling them. Then do a best fit analysis with the known points. This could be done in something like Autodesk Recap.
Thanks for the informative vid. I'm using a SOKKIA ix Robot with Carlson Surveyor+ for survey work. When I loose the prism lock I have to push the "Search" button on RC-PR5A to resume it. How can I use the Data Logger to do it? Also I feel my Carlson Surveyor+ is much slover than your Allegro 2.
Hello and congratulations on the channel, I wanted to ask you if I should recommend a device between iPhone and iPad 11 pro which one would you recommend me to make a survey through a cloud of points with lidar technology? Obviously I am not interested in having a millimeter precision, I know that in this field we need a professional instrumentation. Thank you
The iPhone and other cellular phone positioning systems are pseudo-GNSS systems that receive signals from cellular phone antennas, so they can only be used in urban areas with relatively large numbers of antennas installed. Since cellular phones cannot be used in areas far from urban areas, they cannot be used to build vacation homes or mountain huts, of course.
So... I used the iphone for my side yard, and imported the point cloud into civil 3d and created a surface from the point cloud. The apparent slopes are rather absurd. How do you know which points to select for creating a surface which can be used to calculate possible cut/fill?
Best video, i adore your content. Keep feeding us with ( technology, new equipments of Surveying and process of work for whole your 10 years experience ) much thanks bro
Impressive work : ) I will not nitpick, but the correct way to position the point cloud, is not to align two points. The correct way would be to align for lowest overall distance to reference points. Because of that, the result was actually a bit better than what you ended up with, and sorry if you mentioned this, but I missed it.
The video is excellent. Major snag is the use of some local measuring system which is not universal. Why not use the metric system, the universal system everyone uses except one place?
Im a Geologist, in short I prospect for new valuable mineral deposites, find then, file the claim, if its really good ground I keep the claim and everything else I sell. I was just reading that Geologists are starting to use LiDar to find mineral deposites. Have you heard anything about this? Im wondering if there is a particular type of camera I'd need to do this? Sorry for the questions, I just wanna make sure to get the right onee from the start. Ive been using drones to find mineral deposites for several years really before i seen anyone doing this so if possible i'd like to setup on of our drones with the Lidar
I've been surveying since 2005 and have looked various times for a TH-cam channel to show my friends what I do. They are usually so dry tho and it bores everyone lol. Not thus channel tho. Can't wait to share this with my friends and family! Great video and I will watch some more of yours now. Thank you sir,
Would like to know the undirect comparaison. Like what distance do you shoot between point 7 and 8 with the total station, and what do you have with the iphone. What delta you would have between point 4 and 5 with total station, and what delta you have for those points only with iphone. I'm curious to compare both geometry separatly. Thanks for the video its very interesting !
Great content, as other surveyors already pointed out you should transform with minimum 3 points, preferably with a helmert transformation / smallest square method. Thanks for taking the time making this comparison and thanks for using mm. Regards, a Swedish Surveyor
If I remember correctly, human process and consistency is first error source of any survey. A second run with the total station will have different values. To your point, sensor and software technology has advanced tremendously in the last 20-25 yrs.
Rami... the lidar on the iphone and ipad pro is good for when good enough is OK. ie you want to scan a scene and import it into blender so you can make a virtual world...ie looks good but may not be 100% dimensionally accurate As an architect I wanted to see how accurate the ipad pro lidar was to take accurate scans of existing buildings. I experimented by taking 3 scans of my complex shaped living room and compared the dimension accuracy of the scans to DISTO laser measurements. None of the lidar sans matched the real measured dimensions ... worst case was the 3 scans did not match each other, with a difference of 75mm between them on some measurements... that's 3"... none where close enough to do accurate cad dwgs for a renovation. I have to still survey existing buildings the old way... take laser distance measurements with the DISTO and sketch the plan out on paper showing the measured distance, then draw it in cad program. Its really important that I can trust the measurements taken from the site as being accurate... and plus or minus 1/2" is useless I want/need plus or minus 1mm. To prove to your self how bad it really is to use the iphone lidar. take 3 separate scans of the same area and compare the 3 scans to each other .. in my experience they will not match. and for reference take laser measurements to see how far off each lidar scan is from the actual dimension. I'm experimenting with different photogrammetry software which is looking promising (way better than the apple lidar) especially when I take some actual DISTO measured control points and adjust/scale the photogrammetry mesh so the control points in the mesh are dimensionally accurate (ie set to the actual measured dimension). Ie I measure the over all length of the building and scale the mesh to match.. then the dimensions in the mesh for every thing between the control points get more accurate.
I've a comment on the method of postprocessing if I may: You use 2 points to align the iPhone Lidar points to the Total station, I would have use 3 points to "block" the rotation of the cloud around the line defined with these 2 reference points: A reference frame is defined with at least 3 points Or did I miss something? Cheers
Hi Boris, you are correct, it is a three point alignment it correctly. However, if I do this, it will distort the vertical drift and we won't be able to calculate the vertical drift of the sensor.
@@RamiTamimi achsooo You mean that the vertical is defined with gravity direction, on the total station and iPhone as well. In that case, indeed 2 points are sufficient. Thanks for you answer
Rami excellent work! Just a few quick questions. 1. When Aligning the point cloud to the total station points if you use 2 points (1 plane, X) is that introducing error in the Y plane with 3 points (2 planes) could you balance both X and Y? The Z axis between points looks way closer that the Delta Averages by your chart just wondering if there might be drift between Gyro Level in Phone and total station level. 2. If you scanned say a 1"X1"X1" cube in the LiDAR data would it help with the scaling issues importing to CAD?
Hey Rami, I would like to do a 3D print of a piece of land I owned to better position the future house. The land is sloped and very uneven and is still in a forest state. How do I go about scanning the topography of the land? Thx
Great experiment Rami! Could you share the excel sheet? I think you could further process the data to show other aspects of the accuracy. In particular when I see that all the northing errors are negative, this may suggest that the initial matching of the two datasets may be off (- especially because you used only two manually picked points). It might be better to reposition globally or to look at the error distribution...
I definitely think that Primoz Brglez's suggestion on using a gimbal could increase the accuracy, but i wonder if you change the range from 5m down to 1m (or lower) with a gimbal and scanned the area whether it would be even more accurate again. I think it collecting points from 5m away could be whats creating some of this error? i dont have an iphone to test and know for sure, but just thought it might be something to test as well? Look forward to future videos!
How did you establish the coordinates and elevation of your setup point and backsight point? Assuming you either set two points with GPS or if you wanted to be extra accurate I'm guessing you came off a monument. Although you'd still need a preset setup point if you did the later of those two. Just curious
Surveyor here. I’d love to see if the accuracy improved if you used 3 points to align the scan. You could do this completely in the office without redoing the field work; just use a topo point for the 3rd alignment point (though technically the 3rd should also be a control point). This will likely introduce a little more horizontal error which you COULD adjust for, but I would at least hold the elevation. At least 3 points should always be used when aligning a 3D scan. Imagine a capital letter “ T “. The 2 control points would be at each end of the E/W line segment at the top of the T. By holding only those two points, the N/S line segment of the T can freely rotate, or twist so to speak, around the axis of the E/W segment if the S end of the N/S segment is not controlled. In fact, I’m curious as to how the scan came in even remotely close to being level and not rotated around the axis between your 2 CPs to some crazy extent such as the ground being nearly vertical for example?
For a hands free mapping those are huge results, really good data.
Yes, the results are fascinating for something like this!
For ball park measurements calculating building materials required, this is accurate enough. For scanning between known fixed points, it is also good enough to generate a profile. Given that the maximum range is only 5 metres, you are pushing against the accuracy limits anyway. Pretty impressive performance from the iPhone given it is not mounted on a tripod with a fixed remote reference point.
💯
We tried it and we're off by 8.6 inches over a paltry 61 feet. Worst part was that if we hadn't checked our work with a laser scanner, we wouldn't have known that we were that far off.
If that's "in the ballpark", that's an awfully big ballpark.
@@chrislutes2882 as was said, it’s accuracy is 5m, not 61ft 😂
This is ideal for indoor mapping more than anything, scanning small rooms etc, not for large scale work, that’s not what it’s designed for.
Pretty impressive accuracy for a tiny sensor in consumer hardware. I’m a Geotech engineer. I can think of a few cases where that could be handy.
There is always a great application for every tool. Glad to see more engineers are finding value in this one!
I believe lidar is being considered by nasa for mars exploration
@@RamiTamimi I drive lorries and I sometimes double check the height of a lorry before going under bridge or trough narrow gate and so far have not been dissapointed using measuring tool built in iphone 12 pro max
I’ve been making some crazy architectural diagrams with the LiDAR feature. Had it first on my iPad a year or 2 ago and blew some of my tutors minds back on my degree.
Me, too. Wish I had this thing 20 years ago when we were measuring a route for a new sewage pipe - 10km long, 2m wide, doing it with just a total station and no GPS. That LiDAR thing accuracy would have been way enough and time spent on measuring a fracture of what we took.
This is exactly what I searched! A comparison between a professional and the new iPhone LiDAR sensor.
Awesome, hope you liked it and learned from it~!
I once had to set a bunch of stakes on a property line for my girlfriend…. It was a pretty complicated property line, with odd arcs and angles and set on a steep hillside. I took a plat map of the property and used my iPhone for bearings and a laser ruler to measure starting from a boundary marker. I used a construction app to correct for elevation based upon measured angles to each stake as it went up the hill and calculating the base off the hypotenuse. And double checked with the GPS receiver on the phone for elevation changes. For arcs I set a stake at the center of radius and just measured off the stake to mark out the curves. The neighbor decided to hire a surveyor to check the property line and the surveyor found every one of my stakes to be within an inch of the true property line. He asked me how I did it and was astonished that I could figure it out with nothing but an iPhone and a laser ruler. iPhones are essentially tricorders.
please make a video
@@M-PASTA That was 7 years ago. And I’m not with her anymore…. But even 7 years ago, you could use an iPhone to ascertain bearings and read GPS elevations. I used an app called Theodolite. If you have one known point in the form of a boundary marker, you just measure distances and bearings as listed on the map. The only hard part is correcting for elevation as the plat map is a flat projection. But the property map had topographic lines and numerous bench marks noted for verifying elevation. By averaging the GPS elevation measure with the nearest benchmarked elevation, I could derive a hypotenuse length based upon the plat map distance and measure up the hill that distance ( when you’re doing it on your own, you need a bright surface to target the laser ruler on and you need a laser ruler that can handle the 200 plus feet you’ll need to measure. So you plant a target at a known point and walk the bearing until you get the distance you’re looking for. ) You can double check your elevation estimate by measuring the angle to the prior known target with the iPhone’s inclinometer and deriving how much higher than the target you are with the same Pythagorean calculation. Then you go back down and fetch your target and relocate it to the new known point. My target was literally a length of galvanized electrical conduit with a piece of stiff white board on it with an X at exactly my eye height above a hole with a bolt thru it drilled perpendicular thru the pipe about 18” up the conduit and I would drive the thing into the ground with a short length of larger water pipe slid over the conduit until the bolt I was banging against was at grade- using the iPhone to make sure it was plumb. This way I am measuring from my eye height thru the iPhone to a target that is my eye height above grade. I literally used an improvised plumb bob to transfer the position of the iPhone to the ground. ( holding the upper end of the string in my left hand just below the lens of the iPhone ) As you compile points, If something is significantly off, you run a do over. Over plotting a whole bunch of points, you can tell if you’re on track by the cumulative error. It took a lot of traipsing up and down the hill, resetting stakes and target to correct for cumulative error… but overall errors over and under true measure tended to average out. For example, I would stake maybe 6 points along what the map showed as a straight line… and then sight back along them and correct to the mean line of sight that was on the correct bearing.
I knew I was on track when my plotting placed me within a few inches of a corner marker hidden under a bush that I didn’t even know was there.
All in all, it gave me profound respect for the guys who had surveyed the first ‘accurate’ height of Everest, by starting at sea level in India and working their way across hundreds of miles using nothing but a theodolite and slide rule. I would not use this technique to survey unknown ground, But with the backup of a plat and topographical map with known points you can cross reference, you can plot a property line reasonably well.
This guy is great. A true nerd, but I think we can all relate.
@@calebwallace9589 right?! I just appreciate videos to see their method in action to see if I can find good ideas to apply to my projects.
@@christopherpardell4418 thanks for taking the time to detail and share this, well done mate!
Just wondering, would having the phone attached to a gimbal like the Osmo help with accuracy? I can imagine a certain % of error could be due to shaky hands.
Great video idea!!!
i was thinking the same thing while watching this video.
Buen punto
It might help a little bit but the pull rate of the sensors (gyro, compass, accelerometer) in the phone seem to be high enough to easily calculate out pretty much any shaking in processing. It’s really impressive for such a small device and I wouldn’t have thought that it would actually be that accurate before watching the video
Short answer is no.
Surely you need three alignment points for a 3D dataset, or your elevations will be off for all but the two alignment points and anything directly between them.
Use the metric system and convert to imperial if you want accuracy, not the other way around.
I am impressed with the accuracy of the iPhone-generated point cloud, especially considering the phone was not positioned or moved in a controlled manner. To take it to the next level, the LIDAR app could recognize objects placed on several KNOWN precise control points (as done in the video), the known exact locations of those (2 or more) control points would then be loaded into the app, and the app would apply corrections to all the data based on knowing several exact actual positions within the data. Probably could be worded better. End result would be quite accurate surface data using the iPhone sensor in conjunction with the total station or an RTK system.
It might be possible to use GPS for that.
Great Job Prof. Rami. I am doing some tests with my Iphone13 Pro and 3D scanner app. Only one question: how can I export the point clouds from the app in .rcs format? I can't find this function in the app. Thanks in advance
Did you find the answer to this?
I’m genuinely impressed how many ways you tested this within a 30’ long area of sidewalk and gutter.
As a new real-estate agent in AZ, where so-called "mustang subdivisions" are scattered all over the place, having a strong understanding of surveying only helps our potential buyers to identify risks and possibilities BEFORE they sign a contract...
Also, would love to see same comparison, but using the Matterport Pro2 camera system.
I greatly appreciate your videos, very informative and delivered in a way a layman can understand!
I’ve been in the heavy equipment grading side of construction for over twenty years. The company I work for recently invested into gps technology, I’m the sole operator, I hope to train more, of the gps grader, rover, and I make our maps via Trimble business center. My previous experience with grading was laser with a grade rod. You have helped me tremendously with filling gaps of knowledge I was missing. It’s hard sometimes to ask a question when you don’t even know how ask it, much less even know your missing key knowledge lol. I look at brochures and promotional videos of the equipment we have to grow questions to look up. There’s not much info on TH-cam for this line of work.
I really hope your channel gets traction and gains in popularity, I know it’s a big time suck to make these informative videos but please keep up the great work!
Thanks Kenny. I love making these videos, and having this be my full time job would be a pleasure!
.9 of a foot 🤷♂🤷♂. gotta love continuing US dedication to ft & inches
Thanks for the video. I'm a surveyor from Australia and was intrigued by the LIDAR on these phones.
I'd be curious to see the results if you did scale the point cloud to the surveyed points. The errors look a similar order of magnitude the further from the base point. The error on the elevation though is a bit strange.
Thanks for the metric much appreciated. If you want the table to look neater you can use whole numbers for mm. Only use decimals for super high accuracy surveys.
Jason I’ve found hard stand to be within 50mm and up to 200mm on “grassed” areas, there’s a bit more testing to do and a better methodology to follow but it’s getting there. I’m a surveyor in WA if you want to ask some questions or look at the point cloud
The elevation error could be caused by the AutoCAD align command. When 2 points are used for the ALIGN, the command asks for the scale option. You can also align using 3 points. With 2 points, the point cloud can be slightly skewed from its original horizontal plane, and it is aligned to a line rather than to the plane. A better way would be to average X,Y,Z coordinates for the control points, then do the same for the total station points. This would establish a center of gravity for both datasets. Then you could place them on top of each other using this centre of gravity. Similarly you could calculate average angles for the pairs of points to obtain an average azimuth for both data sets and rotate the point cloud accordingly. Then you could mitigate some of the total station error by taking measurements of all points from various foresight points. This would create an overconstrained network of points, which would allow you to average the errors using Truncated Taylor Series (long story how this is done, but it works magic). This way you would see each control point as 3-dimensional probability ellipsoid with a range of errors, as well as the center of this ellipsoid. Then you would be able to mathematically prove the accuracy of your control point and the point cloud error.
Sorry for the long-winded and academic explanation of this problem. It would make a great project for an advances surveying class.
Great video, very informative and the methods shown were sufficiently accurate! Thanks!
do you have a video just measure a four points property line, and compare GIS, total station and LiDAR?
Hello. Work with measurement of indoor environments for woodworking. In your opinion, would the iphone have the necessary precision to measure a kitchen or a room?
I have an iPhone 12 pro. I would love to just know how to use my LiDAR scanner when I can do with it for every day uses. I tried watching this and I understood what you were doing with the 3-D app on the iPhone but other than that I didn’t understand anything. Especially when you get into the software I had no idea what you were doing. Maybe you can make a separate video and every day uses of the LiDAR scanner. Thank you I found you by your drone videos I am very interested in serving the drones. And my part 1 07.
Great work! I would love to see the efficacy of this on a larger scale. As a landscape architect this could be extremely useful
I know this video is a year old but did you try it with a "better" Lidar capability app like Polycam?
I see a lot of these outside but don't know how they work, thanks for the video, and learning about iPhone LiDAR as well
It appears the local accuracy is better than you have concluded. The errors you are showing in Z are mostly from the registration process being a 2-point method. You will get much higher accuracies if you do a 3-point registration. i.e. the phone's "level" was off from the total station level, which is expected given it's coming from an accelerometer and not a 10" bubble. This put points that lie near the line extension of your two registration points in high accuracy (105, 106, 113, and 114), and points to the upper left in a negative Z error (101, 102), and the points to the lower right with a positive Z error (107-111).
I took your points and did a 3-point registration without scaling, shifted the points for best fit by the average NEZ for all three control points, and then got the standard deviations: 0.10' E, 0.10' N, 0.04' Z, and max absolute errors of 0.19' E, 0.16' N, 0.06' Z. The remaining error was clearly a scaling error as the error directions all radiated from the center.
The fact that the Z had less deviation is because the camera was facing down, perpendicular to the scan surface. If you want better scaling, I think some targets with vertical surfaces placed every 30' would give you that. Maybe try it with some lath stakes at the ends. Ping pong balls also make good photogrammetry targets, and would probably work with lidar as well.
Another good test would be a route between two points and see what the real absolute error looks like. This test really only gets you local errors. That being said, it's looks like it would be a good tool for some applications.
Excellent demonstration. You really did put a lot into this video and it shows and for someone like me that hasn't Really seen a lot of this, it was a very good video and it taught me a lot thank you
Thank you, that means a lot!!!!
What about augmenting a Rod and GPS SENSOR IN COMBO with the IPhone LiDAR?
Dang. Can you use the iphone lidar to create a 3D model of an object and then export to a filament 3D printer? Maybe even scale the object down to fit the print area?
I appreciate your hard work behind the video
I helped my dad do Survey work for our family's construction company back in the late 90s. How I wish we had a little robotic adjustment like that between the rod and station.
good video. Remember that good survey practice is to have your backsight as the furthest point away from your detail survey.... if possible to avoid angle error. Also you had high vertical angle readings, so RL error would be higher. Would be good to see your next video on this.
i used most of lidar scanner in ipad pro it pretty handy to do quick servay and design ..
What errors if you transformed Lidar to 3 survey points? Short rod tough on the knees.. Needs an extension to standing height with the bubble at eyeball level, leaving the reflector low.
I think adding some more 3d features would help improve the height accuracy. Like by leaving some concrete blocks in a grid every 2m. The software likely struggles to tell if mostly flat ground is flat or slightly curving otherwise (when it's stitching together the various 5m sized frames from the video feed to model an area much larger than 5m).
Really nice comparison. This portability and accuracy for visual effects is amazing.
Great, thanks for sharing this comparison,
Amazing and can site exclamation an example thank you!
Please, can you compare the iPhone LiDAR with the Terrestrial LiDAR Scanner, to determine if the iPhone will be appropriate for taken 3D structure or point of a plant. Thanks
I’m a real estate lawyer and actually had an appointed expert-surveyor in a case take a LiDAR scan with his iPad Pro, to scribble notes on and take rudimentary measurements from. And all the (many) parties asked about the app and how he did it. It was so funny.
What do you recommend for a PC to start learning CAD and TBC?
Can you tell me about that receiver sensor? I have never seen one of those and I love it! I would like to use that with my total station.
Thank you for the video. I have all of the corner pins of my property marked. is a string accurate enough to use to set a fence on the 628'boundary line or would you use an instrument of some type?
تجربه حلوة يا رامي التميمي .. تحياتي
Sir I like your Video. Please tell me exporting format from the 3d scanner app. Which type? Plz
Rami , what if you use with Iphone DJI + Posisioning point sticker (which ussualy used for 3d scaner )?
"5.04 feet" - surely if you're going decimal anyway, you may as well just switch to metric?
I would like to see the Lidar sensor with a gimbal and how much it will effect the accuracy
Rami, this is the first video I’ve seen of yours and i found it to be fantastic.
I’m not a surveyor, nor even in the construction industry. However, I do watch a lot of home building, construction, excavation, and concrete videos, so thats probably why your video came into my feed.
This was so well explained I was able to follow and understand the whole process despite never having used either the survey tools nor an iPhone 13.
Here are some thoughts I had.
1.) The Total Station obviously has a lot of advantages afforded to it, for example, it’s fixed position. The biggest advantage, however, I think would be the prism.
2.) While there will be movement induced inaccuracies from the LiDAR sensor and accelerometers in the iPhone, I think there may be resolution and backscatter inaccuracies in the point cloud.
3.) A set of inexpensive markers could be made with a bubble level and some object the LiDAR could accurately key off of (a small orange ping-pong ball, steel ball bearing, 6-sided die? obv. testing needed).
4.) With an effective marker in place, the survey points in the point cloud would jump out and we can subtract the marker’s height from the measured elevation.
In theory, i think this would reduce the prism’s advantages, and some other’s suggestions like using a gimbal or a weighted steady-cam could help reduce motion induced error.
Thanks for the great video! definitely sub’d. 👍🏻
Diddo.
looking to use the iPhone Lidar to capture deflections of a concrete ceiling. Would you be happy to share your excel spreadsheet? I'm hoping to get an accuracy of 2mm or better, do you think that is possible using the iPhone?
I used ti work with Trimble and they don’t need the attachment on the prism . And all in all it’s better with Trimble
Very good analysis. It was helpful, since I have been thinking to buy leica BLK 360, I see for generating rough existing plans with +/- 100mm accuracy this is great. Thanks man.
How did you export the point cloud in "rcs" form? Or how did you change it to that format after exporting in a different form?
how did you convert the point cloud to rcp format? what app did you use?
Perfect intro for a layman. Good analysis between accuracy and precision. Impressed that the IPhone software was able to stitch the cloud together.
Fine for estimations; not suitable for a design survey. Additional control might aid in accuracy.
Keep the land surveyor over the IPhone. But I'm biased.
Wonder how would the error fare on large area.
IMU sensor [which phone use to detect its movement] has inherit errors which would accumulate under prolong use.
Not to be a stickler, but seems like the deltas are somewhat correlated with the direction and distance. Deltas between iphone and TS on local north grid are fairly tight while west and south, deltas are building up. In you check, we can't see in the video if a HChech and VCheck was performed ... Transit and Face to CP. We should not see a correlation between these deltas correct?
Thank you for the video before I study in a college or in a university next year.
For the uninitiated, 0.01’ is about 1/8”, so I gather that the iPhone has about a 1”-2” accuracy for this scan. Probably good enough for estimating take-offs for materials and construction boundaries, but we will probably have to wait for the iPhone 16 before we’re aligning machinery with it. 🤓
Your error could be from skew of the scan. This should be done with picking all the points in the scan coordinate system and labeling them. Then do a best fit analysis with the known points. This could be done in something like Autodesk Recap.
Thanks for the informative vid. I'm using a SOKKIA ix Robot with Carlson Surveyor+ for survey work. When I loose the prism lock I have to push the "Search" button on RC-PR5A to resume it. How can I use the Data Logger to do it? Also I feel my Carlson Surveyor+ is much slover than your Allegro 2.
Hello and congratulations on the channel, I wanted to ask you if I should recommend a device between iPhone and iPad 11 pro which one would you recommend me to make a survey through a cloud of points with lidar technology? Obviously I am not interested in having a millimeter precision, I know that in this field we need a professional instrumentation. Thank you
That's thorough !!
The iPhone and other cellular phone positioning systems are pseudo-GNSS systems that receive signals from cellular phone antennas, so they can only be used in urban areas with relatively large numbers of antennas installed. Since cellular phones cannot be used in areas far from urban areas, they cannot be used to build vacation homes or mountain huts, of course.
So... I used the iphone for my side yard, and imported the point cloud into civil 3d and created a surface from the point cloud. The apparent slopes are rather absurd. How do you know which points to select for creating a surface which can be used to calculate possible cut/fill?
Are this way is suitable for Engineering surveying a large area (for example 100 ha.) ??
I wonder if having a hand held gimble when filming the scan would have a better result? maybe a future vid?
Great work, have you tried an APP called everypoint already?
Can i use the iphone pro for a 3d scanning property scanning business ???
Best video, i adore your content. Keep feeding us with ( technology, new equipments of Surveying and process of work for whole your 10 years experience ) much thanks bro
Great experiment! Thanks!
Thank you! Cheers!
What Lidar app are you using?
Impressive work : )
I will not nitpick, but the correct way to position the point cloud, is not to align two points.
The correct way would be to align for lowest overall distance to reference points.
Because of that, the result was actually a bit better than what you ended up with, and sorry if you mentioned this, but I missed it.
helpful tutorial videos for surveyors.. keep it up.
Thanks for the video, I found it very interesting
Interesting project and discussion. Watching from the Philippines.
The video is excellent. Major snag is the use of some local measuring system which is not universal. Why not use the metric system, the universal system everyone uses except one place?
Im a Geologist, in short I prospect for new valuable mineral deposites, find then, file the claim, if its really good ground I keep the claim and everything else I sell. I was just reading that Geologists are starting to use LiDar to find mineral deposites. Have you heard anything about this? Im wondering if there is a particular type of camera I'd need to do this? Sorry for the questions, I just wanna make sure to get the right onee from the start. Ive been using drones to find mineral deposites for several years really before i seen anyone doing this so if possible i'd like to setup on of our drones with the Lidar
I've been surveying since 2005 and have looked various times for a TH-cam channel to show my friends what I do. They are usually so dry tho and it bores everyone lol. Not thus channel tho. Can't wait to share this with my friends and family! Great video and I will watch some more of yours now.
Thank you sir,
Thank you. Very detailed.
Sir, can you make more tutorials on surveying instruments and methods like leveling, traverse, etc please. :((( thank you
Here you go!
Introduction to Land Surveying
th-cam.com/play/PLH4lR7bcTmFmwqaikJkpoQjrPp4tpA7j6.html
Great video Rami
what are your laptop model and specs?
Thank you. My laptop isn't that powerful, i just use it to remote into my desktop which has 32 GB of RAM and runs on a RTX 3060 graphics card.
Would like to know the undirect comparaison. Like what distance do you shoot between point 7 and 8 with the total station, and what do you have with the iphone. What delta you would have between point 4 and 5 with total station, and what delta you have for those points only with iphone. I'm curious to compare both geometry separatly. Thanks for the video its very interesting !
Great content, as other surveyors already pointed out you should transform with minimum 3 points, preferably with a helmert transformation / smallest square method. Thanks for taking the time making this comparison and thanks for using mm.
Regards, a Swedish Surveyor
Hello, I am a surveyor from Russia. Can you tell us how a typical surveyor's day goes in Sweden?
Dear Sir, i had a question how you save it in .rcs format as I m new to ladar survey in order to download it to civil 3d.
Very interesting test, did you consider using stabilisation gimbal on phone's perforated matrix infrared sensor?
Sounds like an interesting video idea!
If I remember correctly, human process and consistency is first error source of any survey. A second run with the total station will have different values.
To your point, sensor and software technology has advanced tremendously in the last 20-25 yrs.
Well done video. does the iPhone export the pointcloud as a rcs file? or do you do that in another software?
Thank you for the metric values!
No problem!
what size Feet are you using ?? and from what continent ?? 🙄
Has anyone used the Iphone 13 Lidar & 3d Scanner App for archaeology?
Rami... the lidar on the iphone and ipad pro is good for when good enough is OK. ie you want to scan a scene and import it into blender so you can make a virtual world...ie looks good but may not be 100% dimensionally accurate
As an architect I wanted to see how accurate the ipad pro lidar was to take accurate scans of existing buildings. I experimented by taking 3 scans of my complex shaped living room and compared the dimension accuracy of the scans to DISTO laser measurements. None of the lidar sans matched the real measured dimensions ... worst case was the 3 scans did not match each other, with a difference of 75mm between them on some measurements... that's 3"... none where close enough to do accurate cad dwgs for a renovation. I have to still survey existing buildings the old way... take laser distance measurements with the DISTO and sketch the plan out on paper showing the measured distance, then draw it in cad program. Its really important that I can trust the measurements taken from the site as being accurate... and plus or minus 1/2" is useless I want/need plus or minus 1mm.
To prove to your self how bad it really is to use the iphone lidar. take 3 separate scans of the same area and compare the 3 scans to each other .. in my experience they will not match. and for reference take laser measurements to see how far off each lidar scan is from the actual dimension.
I'm experimenting with different photogrammetry software which is looking promising (way better than the apple lidar) especially when I take some actual DISTO measured control points and adjust/scale the photogrammetry mesh so the control points in the mesh are dimensionally accurate (ie set to the actual measured dimension). Ie I measure the over all length of the building and scale the mesh to match.. then the dimensions in the mesh for every thing between the control points get more accurate.
I've a comment on the method of postprocessing if I may:
You use 2 points to align the iPhone Lidar points to the Total station, I would have use 3 points to "block" the rotation of the cloud around the line defined with these 2 reference points: A reference frame is defined with at least 3 points
Or did I miss something?
Cheers
Hi Boris, you are correct, it is a three point alignment it correctly. However, if I do this, it will distort the vertical drift and we won't be able to calculate the vertical drift of the sensor.
@@RamiTamimi achsooo
You mean that the vertical is defined with gravity direction, on the total station and iPhone as well.
In that case, indeed 2 points are sufficient.
Thanks for you answer
@Ramitamimi. Hey, What kind of gears? you're teasing my curiosity :)
Cheers
Rami excellent work! Just a few quick questions.
1. When Aligning the point cloud to the total station points if you use 2 points (1 plane, X) is that introducing error in the Y plane with 3 points (2 planes) could you balance both X and Y? The Z axis between points looks way closer that the Delta Averages by your chart just wondering if there might be drift between Gyro Level in Phone and total station level.
2. If you scanned say a 1"X1"X1" cube in the LiDAR data would it help with the scaling issues importing to CAD?
Hey Rami, I would like to do a 3D print of a piece of land I owned to better position the future house. The land is sloped and very uneven and is still in a forest state. How do I go about scanning the topography of the land? Thx
I imagine the best way would be to utilise a LiDAR drone and then build the model. From there 3d print it.
Rami, this is an excellent video, I'm glad I've found your channel.
Hello Rami, I’m using Civil 3D 2018 but in the ‘Attach Point Cloud’ option in the Insert menu I cannot import .las file that I export with my iPhone 😢
Great experiment Rami! Could you share the excel sheet? I think you could further process the data to show other aspects of the accuracy. In particular when I see that all the northing errors are negative, this may suggest that the initial matching of the two datasets may be off (- especially because you used only two manually picked points). It might be better to reposition globally or to look at the error distribution...
Work such comments am i even a surveyor lol... no wonder i like cars
I definitely think that Primoz Brglez's suggestion on using a gimbal could increase the accuracy, but i wonder if you change the range from 5m down to 1m (or lower) with a gimbal and scanned the area whether it would be even more accurate again. I think it collecting points from 5m away could be whats creating some of this error? i dont have an iphone to test and know for sure, but just thought it might be something to test as well? Look forward to future videos!
How did you establish the coordinates and elevation of your setup point and backsight point? Assuming you either set two points with GPS or if you wanted to be extra accurate I'm guessing you came off a monument. Although you'd still need a preset setup point if you did the later of those two. Just curious
For some projects, it's ok to assume coordinates, direction & elevation.
I’d be curious how this looks if lowered into a storm manhole
The comment I was looking for. I'm tired of fudging inverts with this dumb level rod
Surveyor here. I’d love to see if the accuracy improved if you used 3 points to align the scan. You could do this completely in the office without redoing the field work; just use a topo point for the 3rd alignment point (though technically the 3rd should also be a control point). This will likely introduce a little more horizontal error which you COULD adjust for, but I would at least hold the elevation. At least 3 points should always be used when aligning a 3D scan. Imagine a capital letter “ T “. The 2 control points would be at each end of the E/W line segment at the top of the T. By holding only those two points, the N/S line segment of the T can freely rotate, or twist so to speak, around the axis of the E/W segment if the S end of the N/S segment is not controlled. In fact, I’m curious as to how the scan came in even remotely close to being level and not rotated around the axis between your 2 CPs to some crazy extent such as the ground being nearly vertical for example?
I just received my iPhone 14 Pro. Accuracy question. What would happen if one ran the circuit multiple times and averaged the results?
Starting to use ArcGIS heavy for work and this is exactly the content I was looking for. Subbing for sure.
Thank you sir for your efforts..