Press "Enter" to skip to content

Future Positioning Systems – from Phone to Drone | Fredrik Gustafsson | TEDxNorrköping


evolution has turned humans into a great
innovative explorers and navigators we
are very good at local navigation to
find our way in known environments but
we are not so good at global navigation
I will come back to that the local
navigation all our sensors collaborators
all very complex task that machines and
computers cannot compete with vision is
our primary sensor it builds a local
model of the surroundings in which we
move we support this with prior
knowledge of how such surroundings look
like and contextual information we are
used to in such environments we
complement with force feedback from our
feet when we move and other senses the
inner ear is quite advanced it
completely provides motion and
orientation we have a good sense of what
is up and down our rotation speed and
acceleration and we can use that to date
awake on our movement over short
distances even with our eyes closed that
is we have a redundant sensing system on
a larger scale we use a world model with
the landmarks to navigate ourselves in
known environments this is also a very
complex problem that computers are
struggling with still today but when it
comes to global navigation that is to
find our position in terms of longitude
and latitude we are often lost we need
tools to navigate on large areas to find
our longitude and latitude on the globe
or in a restricted area such as
Manhattan or even in a shopping mall at
Manhattan the way we solve this task
dailies by recognizing landmarks and
comparing to a map mental or physical
map of landmarks that is also something
we need to learn
but they were actually animals that are
superior to us deut evolution for global
navigation like migrating birds and
animals in the sea so this particular
bird was able to navigate to africa and
back again to exactly the same nest as
it left 297 days earlier in loon we
actually computed this path based on
detection on sunrises and sunsets from
the little device on the back on the
board so human evolution was not aimed
for globalization while some animals
actually rolled this ability but we are
inventive to foresee the future we need
to look at the history so the enzian
saleman sailors relied on known
landmarks things as the Northern Star
relates to your heading and also
latitude the Moon and Sun related time
and all the latitude a known landmarks
were used for navigating close the land
comparing to a map and the lighthouses
introduced new landmarks that were easy
to recognize in distance and in darkness
and the light also define the precise
headway the inventive sailors also
located new tools for navigation so maps
have been around for a long time of
course but the invention rate sextant
introduced a way to measure angle
differences very accurately so for
instance that you have the horizon here
and the northern star here that angle
defines your latitude and if you see the
moon and the northwest dog that angle
difference gives you a local time and
time is of course a very important
concept for navigation for instance if
you see the Sun at the highest position
is V die twelve o’clock so you know
local time but you need gain which time
to conclude your longitude
the first generation of clocks used for
sailors it gave a drift of tens of
kilometers every day in longitude but
due to an invention in the mid 18th
century the lift was get used to
drastically and when James Cook to sail
around the globe the final accumulated
arrow was less than 10 kilometers there
are basically three different
mathematical concepts for navigation the
first is the angulation that is to
compute intersection of two or more
lines that can be burying stolen marks
we have a peal activation which is to
compute intersection of two or more
circles which can be the distance to two
landmarks multilateration is more
complicated but suppose that you have to
distance differences that constrains
your position to be on a hyperbolic
function and with two such distance
differences to learn marks you can
compute intersection of two hyperbolic
functions to deduce your position the
ancient sailors used the calipers and
rulers to solve these problems
pneumatically on the map computers do it
today very efficiently computers can for
instance use multiple gadget inspectors
for this task so for instance you can
use directional antennas to compute the
angle the death inspector you can use
time of light or signal strength to
compute the distance to the swifter and
apply the alliteration or you can use
the time difference out to seeing less
on 22 synchronized trance victors to
constrain the position 2 hyperbolic
function basic principles used very
often today the principal invisible all
our navigation tools at land at sea and
in the air rely on satellite navigation
it is based on multilateration
pairing signals from satellites when
they regard to our devices GPS has been
around for a couple of decades you can
find them in almost all of our vehicles
devices and gadgets but unfortunately
GPS has no gundalian see that is if DPS
fails for one reason or another there is
no backup to take over obviously GPS
does not work indoors or underground but
they’re all the situation outdoors when
it fails we have this called Manhattan
problem where high buildings reflect the
radio signals that gives you outliers
and the bed positions I visited
Manhattan lost somewhere and tested
myself and the work I did in one hour
was not reflected by the FPS at all and
I did certainly not walk 30 kilometers
in one hour as indicated here some five
years ago google introduced its position
service it’s based on redundancy just
like the human sensing system at first
hand it uses GPS we’re never available
second hand it uses the closest Wi-Fi
access point that exists in its database
that is used as a landmark in third hand
it applies triangulation based on Cena’s
things from the cellular system so this
is an excellent example of fusion of
different principles using redundancy
and it also brings us to the future of
positioning systems so to get a dundon
see that provides position anywhere and
anytime we need to use many informations
forces combined according to the sensor
fusion principle so we can for instance
have a radio transmitters including GPS
that provides an estimator of absolute
position we have inertial measurements
we making you
inner ear but can be used for dead
reckoning between the landmarks we have
maps and landmark databases to provide
the context of the surroundings and we
have sensors for situational awareness
to make a local map for local navigation
all this information has to be combined
today we have seen a spectacular
demonstration of autonomous vehicles we
have seen demonstration platforms with
multiple antennas for DPS providing
exact orientation of the vehicle we have
yarn base stations providing a
centimeter resolution of position and
etc we have seen the Google car and
Tesla’s diving for long distances out
enormously using short-range sensors
such as grader cameras and laser scanner
but we still have a long way to go to
get a positioning solution that works
anywhere at any time I was smart for
today I have a variety of sensor
modalities and smart positioning systems
are coming in our smartphones but extend
the Google solutions with the new
modalities and the new information
sources so using all these information
needs a framework for how to teach a
computer to navigate and compute the
position the mathematical concepts of
triangulation etc are still very
relevant but we need a context that we
need a framework for how to fuse all
this information and what you see here
is maybe my most famous equation yeah
some of you are seen it before it has
been used by thousands of searches for
positioning applications and we have
used it in several applications as well
so let me show you some domain specific
solutions we have encountered in our
research first one yeah finding your way
in a shopping mall can be a nightmare
you will know that certainly
but you can get a blue dot on your map
by combining sense of fusion of Wi-Fi
access points Bluetooth beacons and
inertial measurements in your phone for
dead reckoning and get exact tech to you
have you move inside the shopping mall
and the solution we developed has been
used in hundreds of shopping malls
around the world including one at
Manhattan what you see here is a replay
on when certain bullbar won the world
championship in sprint orienteering in
Venice a couple of years ago so the gate
line here is what is billed costed live
it’s based on the GPS on his body the
blue line is what was annotated by so a
bull bath after the race and it can be
seen as a young truth that was the way
he actually ran and the headline and the
blue line deviates Venice is not
Manhattan but still high buildings
reflecting the DPS radio signals the
gain line comes from a sensor fusion
algorithm my student implemented that
uses information from GPS and the map
only and it gives the feasible territory
very close to the young truth and much
more worth looking at than the what is
broadcasted today next example shows how
a telephone installed on a drone can be
used to find another telephone hidden on
the ground why yeah the application can
be to find a missing person or a victim
from an avalanche or earthquake assuming
that the telephone is still in operation
so by using the principle of the
litigation of signal strength
measurements on the phone on the drone
the update was able to locate the hidden
phone with a few meters accuracy in less
than one minute so those can be a great
help for localization on the ground and
the next example shows how the thermal
camera can extend the human vision with
the temperature scale and that can be
used for monitoring animals on the
savanna like here
but also defined hidden persons in the
forest which we actually have here and
the application can be to save
endangered animals and protect them from
poaching so to conclude there is no
super sensor for positioning
unfortunately GPS is state-of-the-art
it’s a universal tool today it is good
enough for our dog infotainment
applications but the solution that
provides your position in the time and n
work requires a domain-specific
combination of sensors and prior
knowledge and models and maps i showed
you solutions how to find your way in a
shopping mall improve dps in urban areas
monitor animals and locate missing
people but there are thousands of other
applications out there we have a steady
stream of new application arising all
the time the basic principles on
navigation will be the same but the
challenging challenges for the inventive
human Explorer will continue from phone
to do and further thank you
[Applause] [Music]
Please follow and like us: