hi i'm rami tamimi behind me here is a surveying
total station and this is the iphone 13 pro today we're going to be analyzing the iphone 13
pro's lidar sensor and we will be comparing the data from this lidar sensor to a high accuracy
surveying total station now what we have here is apple's new iphone 13 pro they've included their
second generation lidar sensor next to their three camera sensors that they've all upgraded so
the question is how accurate is this lidar sensor in comparison to a total station a
robotic total station is one of the most accurate pieces of equipment used in the surveying
industry the relative accuracy that we can achieve with a surveying total station is usually between
100th and 200th of a foot so it's safe to assume that this piece of equipment is a great benchmark
to use for testing relative accuracy a total station like this can cost anywhere between 20
to 60 000 us dollars and a professional-grade lidar sensor can easily run you up to 180 000 by
taking several measurements of features using a total station and then extracting those locations
from the data set we get from the lidar sensor we can calculate the relative accuracy of
the iphone 13 pros lidar sensor now if you enjoy this type of content and you want to learn
more about surveying and lidar sensors be sure to subscribe to the youtube channel and turn
on the bell notifications so you're notified anytime we release new videos now for today's
video we're going to be measuring 15 different points and i've marked all these points using a
piece of chalk and i'm going to go over exactly why we're analyzing these points and what we hope
to achieve by measuring these specific locations points number 101 and 102 will be used
to analyze how the iphone's lidar sensor can handle the elevation change between a
curb and a gutter points 103 104 106 107 109 110 and 111 will be used to compare a feature
extraction of the sidewalk we want to see how well the iphone's lidar can position the sidewalk in
comparison to the total station point number 105 is a pk nail a pk nail is generally used in the
surveying industry to set control points so i'm curious to see if the lidar sensor is going to
be able to pick up this feature and whether or not the accuracy of this point using the lidar
sensor is going to be relatively the same as the total station point number 108 is the center of
the storm sewer manhole manholes are really common utility structures that we see in the construction
industry so by scanning it with the lidar sensor i want to see how accurately we can get to the
center of this manhole as well as identifying the rim elevation points number 112 113 114 and
115 will be used to map out the curb and gutter as well as analyze the change in elevation just
like we did with points 101 and 102.
And that's it that is what we're going to be surveying today
this is a very realistic example of what we would do in the industry so by comparing the total
station and the lidar sensor on a site like this we should be getting a real world example
of what to expect in terms of relative accuracy all right let's start by setting up the survey
total station if you're unsure on how to set up a surveying total station check out this video
that i made a while back it kind of goes over the fundamentals of setting up the tripod placing
your total station over a point and leveling it as precise as possible so watch that if you
don't know how to set up a total station and then come back to this video all right and i am
measuring an instrument height of 5.04 5.04 feet okay and i'm going to set my back sight right here
make sure the bubble is in the center and point it in the direction of the total station now i'm
going to take a distance using the total station to the backsight and using this distance i'm
going to then create my local coordinate system okay and we're pointing the total station to the
backsight i'm going to go ahead and take a reading okay and after taking that reading we can see that
the distance between both the total station and the backsight is 24.009 feet and the elevation of
the backsight is 100.289 feet so using these two distances we can now assign coordinates for our
backsight and set up our control network and now we're at the setup menu we are occupying point
number one which is where the total station is instrument height which is the total station's
height which we measured at 5.04 feet backsight point which is point number two and the height
of the backsight is 5.29 feet i'll hit backside and check as you can see we have zero error
which means that our control network is set up i'll hit set angle and now we're ready
to go now i'm going to be using this little 1.4 foot rod this little rod will help
us stay as close as we can to the ground and minimize any kind of errors usually a six
foot rod is fine it really won't create that much error but for the purpose of this experiment
i want to try to do this as clean as possible so that is why we're going to use this 1.45
foot rod for all of our foresights i'm also going to attach this receiver sensor on top of
the prism so that we can utilize the robotic functionalities in this total station and it'll
find the position of this rod without me having to point at it this does not impact the height
of the rod as this is going on top of the prism all right looking good i'm going to start by
setting the rod here on point number 101 okay and the bubble is leveled all right now i'm gonna
have the total station search and find the prism all right perfect we're locked on okay now
that we're locked in on the prism i'm gonna go ahead and assign this as point number one
zero one this is a gutter so i will call it gutter and the prism height is 1.45 feet
and now we can take this observation all right now i've set this up on the
curb so i could just change this to curve and we'll take the next shot and
that's pretty much the procedure i'm going to go ahead and do this for the rest of the
15 points okay point number 103 0.104 0.105 0.106 107 108 109 110 111 112 113 114 and finally 115 and there we go 15 points and as you can see
all of our points are here all of them are also labeled appropriately based on the feature that we
measured i'm just gonna go ahead and finalize this by checking into our backside and making sure
that our horizontal angle is still set to zero okay and measure and we can take a look here and
see that our horizontal angle is zero degrees so we're all set with the total station measurements
now that we're done taking measurements with the total station it's time to test the iphone 13 pros
lidar sensor before we get started i just want to let you guys know that i started surveyshirts.com
here you'll find a large range of surveying apparel the shirts come in different sizes and
colors so please check out surveyshirts.com let's get more survey shirts out there for all of our
fellow surveyors to wear now i'm going to be using the 3d scanner app this app is available for free
on the app store so i'm going to go ahead and open up the app and as you can see here we are in the
low res mode i'm going to switch over to high res and with the high res on i have some options that
i can go through first i can select the confidence level that i want my scan to have the higher the
confidence the more accurate the position of the point is in relation to other points this also
means that there was a strong return on the point so by setting this confidence level to high you're
going to have a lower number of points because the threshold for high accuracy points is a lot higher
so i'm going to set this to high actually because i want to have the most accurate points uh when
doing the scan next we have the range the range will tell us how far out you want the scan to go
if i have a short range it's only going to capture points that are a lot closer to me the more i
increase the range the further points are then collected of course anything after 5 meters will
not be collected i am going to set the range to 5 meters because i want to collect as many points
in our field of view next we have masking which if we're doing modeling work i'll definitely want
to mask the object that i am scanning i don't want the background to show up if i'm doing uh modeling
work but because we're doing mapping work where we're measuring the terrain i want to make sure we
capture everything so i'm going to leave masking off and finally we have resolution the resolution
indicates how much detail we want in our scan if we increase over to 20 millimeters this will
reduce the resolution because now we're scanning at a much higher scale which is why you see the
house there but if i lower the resolution i can set it to five millimeters which would
scan something as detailed as an apple now that i have all my settings in place
i can go ahead and start scanning our site okay okay looking good all right and as you can see we have our scan
right here looks to be pretty good i'm gonna go ahead and process the colorized scan now
all right and there we go the colorized scan is done you can go ahead and turn on
the tin look at that look how incredible that is i love looking at this interface and being
able to see all the connection points for all the points that were scanned using the lidar sensor
now let's take the data from the total station as well as the scan that we got from the iphone
13 pros lidar sensor and analyze it in the office why hello there welcome to my office which is
actually in my house all right let's import all the data into autocad civil 3d we'll start with
the total station points i'm going to go over to insert points from file and i'm going to add
points and i'm going to select my iphone 13 file and there we go the point format is point number
northing easting elevation and description so all of my points are here we'll go ahead and just
click ok and there we go we have all of our points here's point number one where the total
station was point number two where the backsight was and then everything curb gutter curb gutter
curb gutter the sidewalks pk nail and the manhole okay now let's go ahead and import our point cloud
that came from our iphone 13.
I'll come up to the insert tab under point cloud i'll select
attach this is my point cloud file so i'm going to open it now it's going to ask me where i
want to place this point cloud it doesn't really matter where i place it because i'm going to
end up aligning it to the points that i shot with the total station relative path is fine
select insertion point i'm gonna just do 49.50 because 5000 5000 is where the site is located so
i'm going to move it 50 feet off of that site 4950 elevation i'll just set this to 100 because that
was the elevation of our assumed coordinate system the scale will keep it set to one because we don't
want to rescale the point cloud and the rotation will keep it at zero and we'll hit okay and there
we go point cloud is here if i select it you can see the insertion point 4950 4950 elevation 100
zero rotation and scale of one so now when you look at this point cloud you can see we have the
rgb values of all the points the points are kind of small so i'm going to select this and increase
the size to i don't know seven that way when i zoom in i can really see the points so i can see
it right there now it's not the best practice to use the rgb values because there could be a
shift between the lidar points and the camera that was used to take the colored picture because
the lidar points are then assigned colors based on those pictures so if there's any kind of drift
between the two sensors then the colored points are going to be drifted the best way to visualize
this is going to be by setting this over to normal and you can actually see where there are changes
in elevations so what i'm going to do because it's really difficult to pick the points in this
visualization setting i'm going to pick the points with the rgb values but then i'm going to check
the normal visualization to make sure that it's actually where it needs to be so we'll switch
back to scan color and now we're going to align the point cloud to the total station points
it'll ask me to select an object i'm going to select my point cloud and that's going to ask for
the first point so the first point is going to be right here and i see some red so i think that's
where the point is and now i'm going to select 104 now it's going to ask for a second source
point so the second point we said let's do this bottom point here and point number 112 is over
here now it's going to ask me if i want to scale my point cloud and i'm going to say no because i
don't want to change the locations of these points so the relative distances between all these points
will stay the same it's just now they're being aligned to the total station points so that we
can have everything on the same coordinate system so i'm going to say no and that's it the point
cloud is now on the same coordinate system as the total station points all right now we're
going to start the feature extraction from the point cloud that we got out of the iphone 13 pro
to begin i'm going to freeze the total station points so that we're not confused by them when
we go to select the points on the point cloud i created a layer called lidar points so i'm going
to select that and set it as my current layer alright so we're going to go to points point
creation tools and if i bring this over here i can take a look at some of these settings the
biggest one is the default layer we're going to use the current layer and when i come down to
point identification we're going to start with point number 201 all points that are 100 are going
to be points that came out of the total station all points that are 200 will be points that we
extracted from the point cloud that came out of the iphone 13 pro so all this looks good to me i
can close out of that and start creating points so we'll come down here this is our first point and
i can see where there's the red marking so i will select here it'll ask me for a description this
is a gutter and now it's assigned an elevation this is the elevation that it's taking from the
point cloud so we're always going to accept this elevation because that's our measured elevation
from the data set that we have all right if we look over here point number 201 is created right
and i've tilted the point cloud a little bit here so i can at least see the edge of the curve and
select the right point so this right here curb and give it the elevation and this is the procedure
of feature extraction so i'm going to quickly go through the rest of the 15 points and extract all
of them okay point number 103 concrete point 104 concrete point 105 the pk nail point number 106
concrete okay point number 107 concrete point 108 manhole storm sewer point number 109 concrete
point number 110 concrete point 111 concrete point 112 is a curb 113 gutter point number 114 curb
and finally point number 115 gutter all right if we take a look here all the points are in there
i'm going to go ahead and freeze our point cloud these are all the points that we just extracted
from the point cloud i'm going to bring back the point cloud and i want to change the visualization
of the points to normal so that we can take a look and see and make sure everything that we
selected is exactly where it needs to be so now we can highlight the curb gutter much more defined
and it's not relying on the rgb values okay this all looks to be in the correct place yep this
curve and gutter looks good what's really going to show is where there's like intersection
of features like right here where there's a driveway and a curve so this point looks to be
in the correct spot this looks to be the center of the manhole this right here along the sidewalk
and driveway looks good so yeah all the points are matching up to the point cloud like it should
so that means the camera sensor and the lidar sensor calibrated correctly and all the rgb values
are accurately placed so now let's go ahead and export all the points so that we can bring it into
excel and compare all the numbers all right point number 101 which was the first gutter shot that we
took we have a difference in northing of 0.04 ft difference in easting of 0.1 feet and the distance
in elevation 0.22 feet i've also included the millimeter conversions so if you're outside of
the us and you want the metric differences there you go this honestly isn't too bad i feel like
the elevation is a little more than i thought it would be i thought it would be a little less than
that overall positioning wise we're only off by a tenth that's pretty good when we look at the curve
we have slightly more error 0.08 0.14 0.19 in the elevation so yeah again i'm seeing a little more
error than i would have liked point number 103 we are right on in the northing pretty much right on
in the easting and only a tenth off in elevation and i like that that's pretty good that tells me
that um within that distance between 0.104 and 103 there wasn't too much of a difference in there
104 we really can't use to test the accuracy because we used it to do the alignment you might
be wondering why this isn't 100 zero and that's because i didn't pick the exact same location when
i did the alignment versus when i did the feature extraction so forgive me but um i was off by about
a hundredth of a foot nonetheless let's move on to the next point that pk nail was off about
one tenth in the northing and easting and only 300th in the elevation so that's pretty good i'm
pretty satisfied with that one 0.106 which was along the concrete we were off eight hundredths
in the northing right on in the easting and pretty much right on with the elevation so that's pretty
good 107 which is a point on the concrete across from the manhole we have about two tenths in the
northing one tenth in the easting and one tenth in the elevation that's actually pretty good i know
the northing's a little high but um the elevation came out really good on that one so i'm pretty
satisfied the manhole that we shot 108 0.17 0.13 and then about two tenths in the elevation not too
bad actually i think that's pretty good definitely not something i would trust for any kind of design
work because that room elevation really needs to be much more accurate but for what we did it
wasn't too horrible looks like points number 209 210 and 211 share about two to three tenths
difference in the northing 10 to 15 hundredths in the easting and about two to three tenths in
elevation and that's because we're moving about 30 feet away from the start of our scan so i think
that 30 foot move uh did cause a little bit of drift error so i i can see why that would be point
number 112 northing and easting two tenths and one tenth again that's probably from the drift from
how far out we went i'm not going to compare the elevation because we use that point for alignment
points 113 114 and 115 which were the curbs and gutters uh they seem to all share very similar
results two tenths in the northing about one tenth in the easting five hundredths for 115 so that's
pretty good and elevation one to two tenths and so overall my thoughts you know if you're looking
for one to two tenths of accuracy you probably end up finding it there's still drift error that
is occurring with the iphone's lidar sensor so this isn't perfect for mapping however if you
supplement it with some additional control points recognize the margin of error and what your
realistic expectations should be the lidar sensor is relatively accurate it does a pretty
good job of maintaining its positional accuracy and the rgb values are lining up pretty good
with the lidar points so you shouldn't have any issues using the rgb point cloud to do feature
extractions i definitely want to do more testing with this iphone and start incorporating more
control so that we can increase the accuracy as much as possible i hope you guys enjoyed today's
video if you did please like the video i put a lot of effort into this and i really appreciate
you guys watching this all the way through if you enjoy surveying lidar iphones all this
great technology be sure to subscribe to the youtube channel and turn on the bell notification
so you don't miss out on any future uploads check out our private facebook group the link is in the
description we have surveyors from all over the world talking about the exciting projects that
they're working on so definitely be a part of our community and share with us all your surveying
experience be sure to check out surveyshirts.com if you want to buy surveying apparel and
with that i will see you guys next time
