audio Flashcards
Gps accuracy:
dilution of precision:
1/volume of tetrahedron
-4 satellite smallest volume 1.6 almost never get it
Crossprocessing
data collected but not corrected
At 4sigma (standing)
(5) Autonomous - bad 15m
(4) WAAS - 3m
(1) PP GPS less 1cm
(2b) RTK GPS- 1-2cm
(3) NRTK GPS - 2-4 cm
(2a) Opus static
(2c) Opus Rapid Statics
variance in matrix
diagonal of matrix , variance of xyz normalized, always less than 1
time delusion of precision
time error from clock bias
high confidence in position
have small number
if degraded values pick another time of day or not use gps
differential gps
ome place we know the location precisely, set unit there too and get know errors and subtract the know position from computed and have delta xyz to represent error
can be done after taking the measurements called cross precessing or real time processing called real time kinematic processing have broadcast from the station that is sitting on know coordinates to the one that is moving around
known station unit called base the other one rover
we are factoring out the big source of error which is atmosphere, clock bias, position of orbits
WAAS
alk about true form of differential gps
12-16 meter 95 % confidence for regular gps
WAAS (wide area augmentation system) developed TO land automatically plane where 95% confidence limit was 3 meter instead of 12
Station calculation correction facto
here is a central station in west cost and another one in the east coast that communicate with satellite
2 satellite for WAAS one signal west other east coast they are geosincronize, always at the same location in the sky (rotation of earth same as angular rotation of satellite, 23000km elevation above us)
GPS unit set up on top the station in known benchmarks calculate error and broadcasting to satellite , gps unit receives on the main L1 frequency the WAAS signal
cannot survey off this because cannot have that kind of error on foundation of a building
geocentrical orbits
satellite always overhead so angular velocity of earth same as angular velocity of satellite =, moves with earth
Radius dependent, circle earth twice a day
expectation data should be better with WAAS on
Selective availability, degraded signal associsated with gps that is now turn off
Network RTK
broadcast to the service, looks for closest stations and calculate the error the these stations and apply them to you position
What kind of improvement do we get from this?
error reduced to millimeters
gets extremely precise measurement
Opus
e have nova and national geonetic surface of US government set standard for Us and global with collaboration with european unit and japan, set global reference frame for all measurement and update annually
They set up station call Continuous Online Reference Stations (CORS) gps sit units set up permanently, location known to mm, your errors are calculated from this references
OPUS is post processing, they provide the base, errors in cm
problem with height factors out
how does opus compute position
Calculate baseline solution between your position and cors station around you and averages those eros and applies them to your value
need L1 l2 receives, need to collect at least 2 hours of data
Antenna height
supply height of antenna over you bench mark to the reference mark on the antenna
all position get subtracted off give position of benchmark and post processing will take out atmosphere effect and bring us as close as possible to the bench mark value
Negative geoid height
means that he geoid is below ellipsoid
scale factor
mount of distortion you will expect on a map depending on how much the surface of the earth extended about the flat plane of the map
tip to tip valu
latitude/longitude/height value the maximum value of all the data
from hundred of measurement the full range normaly distributed at 4 sigma 95% confidence
Antenna - TPSHGDE
russian and US satellite
Opus good data?
Ephemerous - orbit parameter for satellite, broadcast values 5m of error, 12 hour later ultra rapid solution to try to improve on location of orbits of satellites
Check antenna chosen and height are correct
Peak to peak error less than 5 cm
Which operating station were used
24 hours rapid orbits improve precision location of satellite
2 weeks after,ngs post precise orbits
Opus answer improve with time
how good are orthometric height in opus?
Add error geoid / confidence limit
errir eklipsoid + 5 cm for geoid
ex 4sigma 95% confidence
8cm
Measurement of orthometric height has an added error associated to adjustment
OPUS accurancy is
time dependent
how things improve with time
how long you left gps in one location
longer you keep gps making recording the better the measurement gets
Opus rapid static
session less than 2 hour more than 15 min
baseline more than 3 station
static only 3 station and 5 averaging
for RS can have up to 9 station in less than 250 km
Need L1/L2 instrument
more finicky measurement, can have rejection. After rapid orbits things improve
Accuracy less than static analogous to real time network in terms of accuracy
Normalized rms error by what thy think is a typical day
Cors station organized
backup battey, satellite dish to broadcast , or urban setting internet connection to a service
CORS network in university airport government facility forestation , dense in east and west cost
useful for autonomous driven vehicles specialy rapid static
Range calculated from gps satellite
c*deltaT- clock bias= square root of difference of satellite and user xyz square
730 invention john harrison
problem can calculate longitude on land but not ocean
king George the third offer reward to solve problem
a clock maker comes out with an idea- if we know what time is at prime meridian and what time is where we are, we can calculate longitude, all we need 2 precise clock. One set to greenwich and the other one set to noon and monitor sun and time and set clock and look at time difference
dividing 360 degree by 24 hr every hour is 15 degree of longitude
Laser
stands for light amplification by stimulated emission of radiation
useful making distance measurement, send light beam, does not diverge, known velocity so can calculate delta T and get distance
wavelength and phase relationship
Properties:
monocrmotic
collimated
coherance
beam diversion
an be achieved
We use near-infrared, cannot see light, beyond your visibility, piece of metal aluminum
Lidar
irst arrived in airborne system, simple controllers, against mirror that goes back and ford, put on floor of airplane or helicopter, carried over landscape the oscillating mirror allowed to hit targets that were orthogonal to site
when they got better allow Multi degree of freedom
terrestrial laser scanner placed on tripod
airborne more expensive because of the navigation system that needs to know precisely where the laser is and orientation as it flies
Laser riscan
nit range of 1000 m, but far range measurement are not very useful , most data use if the first few hundred meter
Instrument error of 4mm at 1000 m range
GPS error for ellipsoid height might be 1 to 4 cm the moment we shift our data to an ellipsoid solution it will be degraded
To be an orthometric height then het 5 cm so the number will blow out
We have to register multiple scan together- point do not match perfectly there is some error in there too
we need up we total error at least 1/4 of magnitude greater than actual instrument precision
Instrument are so precise that it is the other things that are coming into place that really matter not the precision of the instrument for most applications
why civil engineering want to use this kind of spacial data ?
structural- look at displacement, deflecttion, rotations, is the construction as built matching the design
More electricty, more piping, new ducts, scan room, see all utility and map them up. figure out what spaces exist
petroleum platform - open and expose and all utility and pumping fluid and electricity and they want to redisign without sshuting down
movie production, taking to set of scene that are going to be shot for movies. Plan out shooting second user
third user mine industry, fast volume calculation
Maptek gear to mining industry in australia
geotecnical engineering soil erosion, any kind deformation associated with land slide storm earthquake
terrestial laser
set up
instrument , camera siting on top that gives it color
above is the differential gps devise that says so many centimeter above reference point of scanner, so we know where our scanner is on earth
set up base station to communicate with rover or use network solution
Laser pointed straight up, Plug mirror and spins, laser comes up and hits the mirror and goes out
run off a laptop and internet connection to base of unit
Out of range that is damaging to eyes
electromagnetic spectrum
Change laser by frequency to achieve different colors
tune to achieved other frequencies
water penetrating green instrument need glasses to protect eyes design to filter out specific frequency
Lidar system
light laser, we have a neodymium-doped yttrium aluminum garnet
oscillator set up between mirror and half mirror and image that the wave bounces out between this two units and leaves the laser and comes out as las light
Energy of light
is a function of a frequency v=f lambda Nd-yag 1064nm given wave length , velocity 3*10^8 v/1064nm =2.8*1064 cycles/sec E=hf=hV/lambda=1.867*10^-19 J energy wave carries
Longer wavelength longer frequency then lower energy, radio wave more energy , microwave
Class lidar laser energy
Riegl z420i
The amount of energy per photon is low enough that there is not risk of being blinded
airborne laser scan, beam stronger and visible an are dangerous
Various type of reflectors that we use to do registrations
one version 10 cm cylinders
retro prysmas directional, need to be looking at it
unidirectional point last at 360 degree and get a return
set total station on a clift put reflector on top of the laser, and some over the total station shot on reflector where the gps unit is on top of scanner to tell us where the scanner is
another strategy is use gps unit to know where the scanner is
porpose the scanner is to just know what is in the land scape but know where the scanner is
set above vegetation, take pic of every set up, so we can unrestnd problems with data
Registration idea of taking multiple scan and bring it torether
two steps translate to another spot and rotation in xyz
to do registration
put reflector out in the field, move scan to different positions, all base line length are different but relation shipp of reflector to each other have not change so we can find the best fit translation and rotation to make them fit in the least square sense
need to collect multiple scan to fill in shadows, make surface, counter map and change detection
Problem with tripods
the scanner can only see what you can see, so vegetation can be a problem, ways to improve is, one way is to try to make an quasi airborne system by installing television poles, can have downward view of terrain, they are called space needles
cylinder reflector
has a height and diameter, when scanned you get to the outer surface of cylinder vand there is an alogarism thatt calculate the center height go the cylinder and subtracts that from the closest distance to that cylinders to determine location
scanner sits in its own coordinate system, does not know where it is, thinks is 0,0,0 , so everything that measure is relative to 0,0,0
two scans, hold one as a reference and the other one freely moves to best match the reference
Coordinates system
All individual scans that are at 0,0,0 but are pull away to a relative position from that landscape , this is called a project coordinate system
Internally is all consistent to the square best fit model but does not know where it is in earth
Last step is to translate it to global coordinating system where we take values and translating to northing and easting and elevations, typically in UTM
Riscanpro
closer than two meter does not show up
define reflector circle with 3800 points
need minimum 3 reflector to triangulate location
camera to color scan
use for utilities, can show hanging light, pipeline etc can be taken to i-studio and figure out how to bring new utility to the room
movie set figure out everything you gonna do in the room before you get there
Maximum rmsc error of half a meter,
In the field take gps unit, take measure of scanner and reflector, northing easting and elevation, select list as global coordinate system list and do best fit to global coordinates
euler and reflectors
n scan did translation and rotation with direction cosine matrix which is a matrix form of euler equation, and found the common best fit set of reflector between scan, sastify the least error
euler angles
we do slight rotation, the sign of the sine angle will be nearly zero, cosine slightly less than 1, x’ almost entirely associated with cosine x and y’ associated with cosine of y
45 degree cosine and sine equal to each other, new x position being a composite of both x and y coordinate equally, part of the x becomes the new y component and part of original y component becalms part of the new x component
the simple translation is cd, once we add third corner of rotation we have a problem, the order of the transformation matters because you get different results. c
turn sine into cosines
turn sine into cosine by cosine x’ relative to x and y component times the sine of alpha turn into the cosine of x’y
turn matrix into cosines calculating orthogonal rotated vector
to do rotation in 3d
rotating about z value xy and rotate nothing happens to z values
map out error differencing position one fix other relative moving passing we can find minimum and that is the one we accept as best
point clouds
look elevation in a model, calculate hight for scans, square of the difference sum up over all increamental measurement N , and take square root of that and is the root mean square error of elevation
calculates residual -difference between the two
Minimum for mean square error — best fit solution
expenses way to do it —translate steps and then rotate
other way determine max vector at rmse is decreasing follow the path into center of ball, much faster
What components causes liar to work faster today than an instrument used in class?
the one in class only 4 years old, but slower, less memory, less speed. transition is faster, step remotes one sponge mirror vertical orientation, stoping and starting, a second rotating 360 degree turn. To do it currently has to happen in a speed that is not too fast, there have been improvement at the speed remotes can do this accurately and stop and start faster. Only real difference btw the one in class and new one is speed
is it possible to scan the entire are in riscan and do registration does it mean we don’t need to register each scan independently in maptek?
re register in lab to try to find the best model
what were the sticky disk in the room?
all those reflectors are registration points, instead of doing global matching in maptek, we match up the reflectors.
Least square best fit
omputed sum of square residual between measure value and known value
Euler angles order
the order by which we do the transformation, will change the orientation.
transformation happening across from position 1 to intermediate position f’ to position f2’ then to position 2
T21 = T2F’’ TF’‘F’ TF’1
how does least square intercepts euler
scan does not know how much it has to move so we have to use step progression to get it into place
taking one scan registering to a second scan t though this matrix manipulation and match them program alogarism and if fail or calculate a rmsv lower so going in the one dicrectio, we keep doing that until we find the minimum
2 scan register together, so we move from individual coordinate system to a project coordinate system, we call this a soft matrix, and then move to a global coordinate system
Comparison between airborne and terrestrial
Airborne came first, step remoter were slow and difficult to do angular rotation,
laser scanner using tilted mirror has bee used for a long time but in this case we are measuring the time of flight of laser hitting the mirror, going to a target on the ground and coming back hitting the mirror and coming back to receiver
terrestrial approach is the demonstrated in class
advantage ground,
stationary - if you want to know coordinates in landscape is better to make measurement from stationary position
if you are close to the scan which mean you can put many laser point onto very small target close up
Advantage air
cover a lot of landscape and fast
but resolution is low and accuracy is also low
trench data set, comparison
airborne and terrestrial
first scan from a plane with window floor
and the trenching was done with helicopter to get close to the ground
the laser range finder 2-3 cm, gps error 5-10 cm, imu error 9cm @ 900 m
imu error
is one of the complication of having airborne liar data, if there is wind airplane has to flyght into it to compensate, so they flight skew pathway changing orientation, wind not always coming horizontal can come upward or downward, so airplane can be pitch up or down, turn oblique or roll. These motion complicating navigation of airplane changing orientation of the way the last moves and therefore changes the location of where the laser hits the ground
immersial motion unit (imu) make really precise measurement s of roll pitching yag to reposition point accurately on ground
IMU Many grades of quality, most acurrent systems are use for defense, ex. jet.
Add up error vendors giving a 95% confidential error of 18 cm vertical 30-40 cm for horizontal (not so great for CE applications)
airborne systems collection
swaths of data, a little of overlap, error increase as we move away from beneath aircraft out to outer extend of scan, so want to have sufficient amount of overlap.
laser does a z pattern coming together at end scan and then spread apart . big gap at tail and fit closer together toward center. we try to exclude the data toward outer edges
airborne benefit vegetation
airborne benefit is that uses different type of algorithm to calculate target. For our purposes we have the speed of light times time of flight over two as range between laser and ground it hits
airborne system apply optics to allow laser points span several meters called beam divergence. More complicated analysis of the return waveforms, not picking up the stronger signal but the whole return wave form signal. Useful for vegetation problem by divergent the beam, it hits top of tree and sends strongest signal, and the other option of laser continuous its way down until it makes it to the ground. hope last return is the ground.
There is a lot of processing that goes into the analysis of the last return to try to find our which one is the ground
hurricane isabel, north carolina
North caroline barrier beach,
Before hurricane, barrier island, homes look degraded in airborne system, do not retain shape
Hurricane reaches though the barrier and some of the structures were lost
cal campus unavco open topography
Open topography to access data,
Cal campus poorly define, looks fussy and unclear, but for a general topographic of berkeley takes 2 min to collect, so you can cover a lot of terrain quickly but not vey accurently
trench study scan
Airplane -poor spot density
helicopter laserr points is better but can’t hardly get a sense that there is a trench there
terrestial data scanner, there is all the detail
trench study stats
know values, standard to compare airborne is total station made by surveyors
residual calculation between known elevation and measure by lidar
Mean of distribution offsets from know value
Mean of the residual is a measure of sample bias, offset between red and blue line
dispersion how spread out from the central mean of the bias, measure of noise of data
Bias measure if there is a systematic problem in determining elevation
4sigma -+2 95 % confidence limit, expect Mui to fall between these two point
sample desperation difference between residual minus bias over all data
root mean square error
difference in residual over n
large sample equal bias square minus dispersion squared
unbiased population equal to the standard deviation of residual or dispersion
One of the problem in airborne system
hat you do not own the devise someone does data collection for you
for terrestial you can control and revisit if there was a problem
poor quality control in determining bias and dispersion between vendors in airborne so difference between 23 cm and 8 cm might be two different vendor and their systematic problems
education aspect of liar
ability to share—virtual trip to a location
3 examples from different event
land slide- hitachinaka japan
structural damage, tilting -chile
breatching damn - japan fujinuma
Not collecting data in optimal way, without reflectors
hitachinaka port japan
overlay port on google earth-road back of port became deformed and depressed in outer edge, and elevation of ground behind structure.
evidence of tsunami, scatter debris, not first floor, no walls
step up at latter skirt, buldging toe of land slide
using section ttool maptek,
see cliff no damage
head skirt-settlement
deformation of toe
no data set before earth quake so we can project surface of the road, model using polygons and use this to calculate settlement
settlement extended from 0 to 2 meters
Concepcion chile 2010 earthquake
liquefaction
one building experience a lot of settlement on foundation
two building were build in one year in a densified mat and two of them build the following year on an not as well prepare foundation
first two building no deformation but the other two 2 cm below
President condominium
Worst buidling we have one corner 34 cm settlement 6.9 on the other corner so its settling and tilting over a foot
3.8-10cm settement on rest building
how do we know settlement? using a loop in maptek of one elevation and drop down to calculate the change of elevation of the corner of buildings
returned to the site in 2013 and can see the building leaning
1.5 degree tilting
2014 two buildings were demolish
fujishima japan — dam failure
dam failure
agriculture dam, earthquake liquify foundation of dam and the up reservoir slid to reservoir causing after to overtop the dam, receiver had fully drained hours after earthquake
1949 dam, performed well given the age
pre event contour drawing comes based on engineering drawings
post event geometry comes from using polygons
one zone of very dense data, obtain by doing overview scan first, zoom in to get a higher resolution scan to calculate post event damage
section of dam that slid into receiver was on the order of 8m and then that left the receiver thought the gap was 8-10 meter more
refraction
refraction is the speed of light in a vacuum divided the speed of light in the atmosphere
very small change
3 ten thousands but has effect on measurements
laser passing though air, velocity drops frequency remains the same so wavelength changes
refraction of speed of light on vacuum divided by speed of light in air
aspects that affect refraction
atmospheric pressure -vacuum at normal atmosphere at see level causing being of light
move from high pressure moving down expect index of refraction to fall alls to zero when get to vacuum
temperature - the wavelength shorten and volesty falls as the temperature gets colder (thicker air)
temperature has the greatest effect
effect of humidity is minor
temperature is controling changes of refraction
refraction for airborne and terrestrial
airborne data set, air temperature is different from temperature of ground and laser light is going through the whole range of temperatures, if there is a real temperature difference we have an uncontrolled error on our range estimate
refraction an issue obliques shots for airplane because the pathway is longer
Also another reason why in terrestrial we like data to overlap and in we use the closest data and exclude other
one of the issue associated with refraction is
that affect the precision by which we can estimate the location on the ground. Refraction changes the velocity and wavelength and therefore changes our estimate of the range and is linearly proportional to the length of the shot. so the longer the shot the more accumulated errors can occur
when doing laser scan need to record temperature, and try to minimize the difference between your set up location and the target you are shutting to shorten the distance to minimize effect of refraction