Sensor Fusion on Android Devices: A Revolution in Motion Processing

Поделиться
HTML-код
  • Опубликовано: 3 ноя 2024

Комментарии • 139

  • @shmojo_jojo
    @shmojo_jojo Год назад +27

    Even 12 years later, it seems to be the most comprehensive talk about sensors and sensor fusion I have ever watched. Thank you!

  • @sanjivsoni2389
    @sanjivsoni2389 7 лет назад +11

    I saw this video again after three months and theres still so much to learn about this. This is so far the best resource on Motion Sensors and sensor fusion I have come across.

  • @Gunner3210
    @Gunner3210 13 лет назад +8

    Wow, we designed the exact same thing to use for our own sensor fusion algorithm.
    Never knew that Google was using something exactly the same. Talk about parallel innovation!

  • @saxman58b
    @saxman58b 8 лет назад +3

    The best explanation of how to interpret quaternions I've seen. Thank you!

  • @miojovege
    @miojovege 4 года назад +1

    10 years Later. Still Amazing

  • @invensense1
    @invensense1 14 лет назад +2

    Edit: In the slide on "double integration", the total time of those data sets are actually 10 seconds, not 1 second. --- David Sachs

  • @dhairyamittal940
    @dhairyamittal940 Год назад +1

    Apple taking notes... "Ripped out the headphone jack"

  • @seppl5372
    @seppl5372 4 года назад +2

    how accurate is the video these days? did the sensor accuracy develop and it works better today?

  • @harleydavidson845
    @harleydavidson845 11 лет назад +3

    Checkout DARPA’s Micro-PNT (Micro-Technology for Positioning, Navigation and Timing) positioning tracking technology, it does absolute position tracking on a single chip with sensor fusion.
    To oversimplify it; Micro-PNT adds integrates a highly-accurate master clock (TIMU) to existing chips gyroscope / accelerometer to simultaneous measure the motion tracked and combines that with timing from the syncronixed clock, and with sensor fusion makes single chip that does absolute position tracking.

    • @ishkool8664
      @ishkool8664 4 года назад

      so did they achieve success?

  • @marquesghm
    @marquesghm 2 года назад

    One of the best talks I have ever seen! 👏

  •  6 месяцев назад +1

    You know the video will be interesting when it is 4:3

  • @AshNapToys
    @AshNapToys 11 лет назад +2

    I'm pretty sure that the gravity bias graph at 25 mins is wrong. If you had no gravity compensation the most you could move in a second would be 0.5g which is about 5 meters...

  • @ApocolypseZombie
    @ApocolypseZombie 10 лет назад +5

    Awesome talk, even from a general sensor overview perspective

  • @avibank5115
    @avibank5115 9 лет назад +26

    This guy is so cool

    • @avibank5115
      @avibank5115 8 лет назад +19

      Lol, found this video again, and was about to comment the same thing

  • @chiubobo4955
    @chiubobo4955 Год назад +1

    I bought an IMU for practing embedded system and try to use integration to get the sensor displacement.
    It's turns out I'm a fool in the perspective of 12 years ago video.

  • @icecream296
    @icecream296 8 лет назад +1

    Can someone explain how the pan change worked at 38:00 assuming I have got the rotationMatrix from the sensor library, how to I pan the 2D image on my mobile screen?

  • @Hasan-s3q6f
    @Hasan-s3q6f Месяц назад

    How can we remove gravity effect from raw accelerometer data? I mean how we get linear acceleration? I have seen free acceleration also in literature. Linear acceleration and free acceleration is the same?

  • @s4ril
    @s4ril 8 лет назад +3

    It was incredible. This video help me a lot to approach my thesis based on staft related with acc. Thanks. GreatJob

  • @belm0_
    @belm0_ 5 лет назад

    Regarding the gravity signal, he says "it's actually mostly gyroscope data, but with the accelerometer used to correct drift" (at 17:59). I can't find any information on how this is done.
    The W3C Motion Sensors Explainer hints at the same thing: "On most sensor hubs, gravity is isolated from the accelerometer using the gyroscope" (www.w3.org/TR/motion-sensors/#relative-orientation), but also lacks any detail.

  • @ASelimSalman
    @ASelimSalman 12 лет назад

    I have proposed a similar topic since Sept. 2009 at my MSc. I wish I had the opportunity to continue researching about!

  • @sarnathk1946
    @sarnathk1946 3 года назад

    Thats nerdy stuff. The guy presented so well...

  • @PatLane4069
    @PatLane4069 2 месяца назад

    Really, really excellent presentation. Thanks heaps.

  • @Tadesan
    @Tadesan 7 лет назад +1

    5:00 street view is run by the same technology pioneered by the video game Mist.

  • @tomp2309
    @tomp2309 6 лет назад +1

    very good video, helps me so much writing my master thesis. additionally he has so a clear pronunciation

    • @BBorn223
      @BBorn223 6 лет назад

      sorry, I know this is inappropriate, but can I see your thesis?

  • @gnychis
    @gnychis 13 лет назад

    great description of why position using the sensors is challenging, thank you

  • @xpeace77
    @xpeace77 9 лет назад +3

    Are the presentation slides available for download ?

  • @gggavin
    @gggavin 9 лет назад

    How would I go about coding an application that outputs "roll" every time the teapot is rotated at a full 360 degrees, and not recording all other motions

  • @kylemichaelbecker
    @kylemichaelbecker 13 лет назад

    @ the question about a dynamic x and y reading, with a static z value, my first guess would be that x and y would reflect lat, long; wheras your altitude(z) may be fairly constant. hope this sheds light

  • @johnkruebbe4276
    @johnkruebbe4276 9 лет назад

    This is how we could do position in VSN. Thank you!

  • @nyehnyi
    @nyehnyi 13 лет назад

    the best online lecture i ever watch..thank you so much sir

  • @ArmendKrasniqi
    @ArmendKrasniqi 13 лет назад +1

    I have Samsung Galaxy S2 and I'm experiencing with an issue that's annoying me very much.
    When i do "auto rotate" , the rotation goes after 4-5 seconds.
    I've done the calibrations and all that stuffs that you will find on google or forums.
    Also, i've checked secret code *#0*# , and at Accelerometer Sensor i get:
    ACC Raw Data - X: changing numbers, y: changing numbers, z: 1024 ( z is not changing like X and Y)
    In this case, is my Accelerometer Sensor broke Physicaly or what?! ?!

  • @syednaim2893
    @syednaim2893 10 лет назад +10

    Much impressive,
    Very informational,
    Wow

    • @Tadesan
      @Tadesan 6 лет назад

      So comment!

    • @tim_meister
      @tim_meister 6 лет назад

      you can really tell this comment is three years old

  • @TheiLame
    @TheiLame 8 лет назад +5

    This is 5 years old video, is it still up to date? :) anyone knows?

    • @saxman58b
      @saxman58b 8 лет назад +9

      +TheiLame The math and physics all still makes sense. What has changed is that most phones now have all of this hardware built in, and on the iPhone at least (I only do IOS development) the software functions are now built into the standard libraries. So no need to hack the OS or add hardware to get the gravity vector, user acceleration, attitude (via Euler, quaternion and rotation angles) etc. It's almost all there and ready to use with little or no extra processing.

    • @axelriet
      @axelriet 4 года назад

      Euler is 237 years old (in 2020) and he's still pretty much up-to-date yes ;-)

  • @HAWXLEADER
    @HAWXLEADER 8 лет назад

    yet still the nexus 6p has tons of problems with its sensors. some gyro axis are not on zero when static, gravity value of the linear acceleration virtual sensor is off and there is no way to calibrate any of them...

  • @Subhareact_01
    @Subhareact_01 5 лет назад

    Hello, my mobile's gyroscope is working delay . How to solve this?

  • @chrisanderson1513
    @chrisanderson1513 3 года назад +2

    "we ripped out the headphone jack"
    This was all your fault. Apple watched this video and this is how it all started.

  •  9 лет назад

    at 24:24 he said it wrong. It drifts only 2cm not 20cm. As you can see later on next graph blue line is on 8,5 but green line is flat, and on first graph is at 0.020, not 0.20. If the error was 20cm then it would be obvious not flat. Or, he maybe thought 20cm after 10sec

  • @bigboateng2011
    @bigboateng2011 9 лет назад +3

    This video helped in my project. Omg thanks!!

  • @kokopelli314
    @kokopelli314 14 лет назад

    Very well done. Simple and concise!

  • @axelriet
    @axelriet 4 года назад

    Interesting how Anglophones call a gyrometer a "giroscope". A scope is normally something you look into and/or aim with. Microscope, telescope, periscope, endoscope, borescope... Meter means "measure" as in thermometer, dynamometer, chronometer, accelerometer, magnetometer and yes, gyrometer that measures angular velocity aka the rate of angular change, from which one can compute the rotational speed. A small mystery of science and technology :-) Maybe there was a gyro-stabilized scope someday, probably invented by Leonaro da Vinci? Modern technologies start to correct that discrepancy: docs.microsoft.com/en-us/uwp/api/windows.devices.sensors.gyrometer?view=winrt-18362

  • @FasihRana
    @FasihRana 12 лет назад

    It is there in Android SDK. Not sure about other platforms.

  • @ProgramMax
    @ProgramMax 14 лет назад

    @ballystix
    Thank you for the response. Unfortunately, I don't quite understand what you are telling me. Is there a friendly website to help me learn what you mean? I'll hit up wikipedia and try to piece it together. But if you have a nice tutorial or the likes I would really appreciate it.

  • @shekhar2345678
    @shekhar2345678 3 года назад

    thank you..it was great explanation with demo

  • @stivio00
    @stivio00 9 лет назад

    Really Really Helpfull.
    Thanks David

  • @ProgramMax
    @ProgramMax 14 лет назад

    This talk is amazing. You were able to take a difficult problem and explain its solution in a simple way. Moreover, you didn't do it by dumbing it down.
    I would like to learn more. For example, I understand how the gyro can be used to assist in accelerometer readings. I get that we no longer need a low pass filter. But I don't understand HOW to do that. Is there a link somewhere explaining that process? Also I don't yet understand why the compass is so important. To help us find down (gravity)?

  • @roidroid
    @roidroid 12 лет назад

    it's an Accelerometer. "Gee" Force (or G-Force) is a measure of acceleration, 1 Gee is 9.8m/s/s (Earth's gravity).

  • @riddlebeats9251
    @riddlebeats9251 3 года назад

    lmao the headphone jack removal years ahead of its time

  • @WarmWeatherGuy
    @WarmWeatherGuy 2 года назад

    23:22 You aren't removing gravity, you're adding gravity. Accelerometers can't measure gravity. When you're sitting still on a table (for example) you're not accelerating. The accelerometer measures the force of the table pushing up but not the gravity pushing down.Once you add the gravity that the accelerometer can't sense you end up with zero for acceleration.

  • @rajatgupta8854
    @rajatgupta8854 Год назад

    VR in 2010, well that was something!

  • @ignoram9us
    @ignoram9us 10 лет назад

    This is brilliant!

  • @DrTune
    @DrTune 10 лет назад

    Exceedingly interesting talk, thanks!

  • @HassanSelim0
    @HassanSelim0 11 лет назад

    I'm here because of something a little bit older, I wanted to get position from the sensors on some Vuzix AR glasses I have :)

  • @kayumiy
    @kayumiy 7 лет назад

    what is the data format of 9-axis combination data?

  • @TheiLame
    @TheiLame 8 лет назад

    42:17 Hi im little bit dumb. How does he get 9 numbers from 3 axis points? How does the twist work to give 6 extra numbers? "Those axis twist around with the teapot..." i dont get it.

    • @saxman58b
      @saxman58b 8 лет назад +2

      +TheiLame I believe what he meant was that each of the three rows of the 3x3 matrix are the vectors describing the new x, y and z axis relative to the original xyz dimensions. For example, if you wanted to just rotate around the z axis by 90 degrees then the matrix to convert old dimensions to new would be:
      [0 -1 0 (vector describing the old x axis relative to the new coordinate system after rotation, i.e. x' = -y)
      1 0 0 (vector describing the old y axis relative to the new coordinate system after rotation, i.e. y' = x)
      0 0 1] (Z axis doesn't change so all vectors keep their same z coordinate)
      To test it, pick an arbitrary position in space, for example (1, 2, 3). If you multiply the matrix above by this vector you get a resultant vector of (-2, 1, 3), which is exactly what was intended, a 90 degree rotation around the z axis.
      This was a simple example, but it should work for any 3D rotation. Just make each row of the rotation matrix equal the vector describing the new direction of the three local x, y, z axis of the object. And unit vectors are required in the matrix, or else the space / object will expand or contract. Thus a 45 degree rotation version of the above would be:
      [sqrt(1/2) -sqrt(1/2) 0
      sqrt(1/2) sqrt(1/2) 0
      0 0 1]

    • @TheiLame
      @TheiLame 8 лет назад

      saxman58b
      Thank you very very much for explaining! :)

    • @markchinski5071
      @markchinski5071 5 лет назад

      Search youtube 'linear algebra matrix transformations' for more info.

  • @artiny16
    @artiny16 8 лет назад

    26:44 I doesnt get it ...why is use Kalman filter, and what was about pedometer. Someone can me explain it ,please?

    • @Ateeb0Malik
      @Ateeb0Malik 7 лет назад +1

      it's for the positioning of the character..for example if you want to
      move your character by moving your device or want to map your footsteps
      as input to your character's step... he was actually giving alternate to
      the Accelerometer (to avoid double integer errors) and proposed
      pedometer and Kalman filter to use for the above mention task.

  • @F60Stunt
    @F60Stunt 3 года назад

    very useful thanks

  • @justinroughley5958
    @justinroughley5958 9 лет назад

    Really useful - thanks

  • @8VT0
    @8VT0 5 лет назад

    Very nice.

  • @seanpe
    @seanpe 11 лет назад +1

    Yeah, the number is wrong. I'm sure he meant 10s instead of 1s. A 1 deg error is 8.5m after 10s. His point stands though, gravity is big and double integration sucks.

  • @daniyalIbrahim15
    @daniyalIbrahim15 2 года назад

    Who is this guy? Impressive

  • @Sentilon
    @Sentilon 13 лет назад

    is anywhere the current presentation to download?

  • @MrKarln
    @MrKarln 10 лет назад +5

    25:30: The drift is 8.5 centimeters in one second, not 8.5 meters. If i drop the device it will fall (displace) 4.91 meters in one second. The integral of a small error on gravity is not larger than the integral of gravity itself.

  • @CurtHowland
    @CurtHowland 14 лет назад

    Thank you.

  • @AleksandarKospenda
    @AleksandarKospenda 8 лет назад +24

    What we ripped out was the headphone jack. Apple 7 copy pasta.

    • @iMpacTbySkil
      @iMpacTbySkil 8 лет назад

      I was looking for this ^^

    • @axelriet
      @axelriet 4 года назад

      Do you realize he said that some 6 years before the iPhone 7? At the date this video was recorded Apple had just released the iPhone 4.

  • @wisehomo
    @wisehomo 2 года назад

    The most important of all is the sensor fusion that was never talked about. lol You showed how it works from application perspective but why it works never explained by going over how sensor fusion algorithms work.

  • @drpolishmatt
    @drpolishmatt 8 лет назад

    Thank you!

  • @Techn0man1ac
    @Techn0man1ac 10 лет назад

    Молодец, знает о чём говорит.

  • @cavins16
    @cavins16 14 лет назад

    ooh all of the big words and stuff make stuff look cool

  • @BurakCalik
    @BurakCalik 13 лет назад

    Clever!

  • @Troysteele
    @Troysteele 10 лет назад

    very informative

  • @iradeourum
    @iradeourum 7 лет назад

    what is the "inegral" ??

  • @iyoutome
    @iyoutome 6 лет назад

    Check out Ken Wheeler "Missing Secrets of Magnetism" 3rd edition.

  • @ballystix
    @ballystix 14 лет назад

    @ProgramMax
    I believe you would need to write out the filter as a bunch of difference equations (which is basiclly the discrete form of a differential equation describing the filter response).... alternatively you can try using fourier/s-domain analysis but i doubt that can done easily on a device like that.

  • @pylaochos
    @pylaochos 9 лет назад +4

    Euler was Swiss, this is why his name is pronouced "Oiler".

  • @ramonalisa
    @ramonalisa 14 лет назад

    euler is pronounced oiler because it is a german word. eu makes the sound oi. kind of like the strange combinations in english such as oo, ew, aw, au, ou, ow, etc....you get the picture.

  • @solomon2571
    @solomon2571 12 лет назад

    What is g-sensor then

  • @AskAFreemason
    @AskAFreemason 12 лет назад

    ... or just use a combination to fix that problem.

  • @canvin
    @canvin 10 лет назад

    Thanks really helpull

  • @Klem98
    @Klem98 13 лет назад

    Hi , where can i find an api to use this sensor fusion algorithm ?

    • @axelriet
      @axelriet 4 года назад

      Lookup TDK/InvenSense MotionApps™ Platform.

  • @TomFoolery9001
    @TomFoolery9001 11 лет назад

    Ha ha yep, I was just thinking that the Oculus Rift guys need to hire this guy and throw some sensor fusion in their new headset.

  • @PietervandenHombergh
    @PietervandenHombergh 10 лет назад

    Nice presentation, useful

  • @charlesaydin2966
    @charlesaydin2966 4 года назад

    Euler is pronounced "Oiler", because it's a German name. :)
    Regardless, GREAT talk!

  •  12 лет назад

    9-axis - it integrates all 3 types of sensors. This can't be 9DOF because you can't change your orientation for example in magnetic field without changing your position or orientation in world space.

  • @brucesha7901
    @brucesha7901 9 лет назад +1

    youtube(google) should integrate IQ and Knowledge into thumbs-up/down. because that way when you see the ratio you'll have a better chance of correctly evaluate yourself!

  • @moiquiregardevideo
    @moiquiregardevideo 7 лет назад

    Interesting to compare the real world design vs the point of view of abstract University teacher who wrote the wikipedia pages about what should be the same subject.
    How the GPS work so well in any cell phone? The real data from the satellite is not precise enough to detect that you arrive at a street in 50 feet. But try to discover how the cell phone work from any web site or youtube video..
    Nobody knows. Wikipedia just mention the expensive equipment that are mandatory in commercial airplanes. They know the cell phone in their pocket appear to match the performance of these devices. But all they can say is : thanks to Einstein, science works. The black hole in their knowledge is discretely ducted under the rug.
    This video published in August 2010 explain the difficulty of sensing linear displacement from the 6 sensors. A good engineer can fill the gaps, extrapolate what the device do starting from the noisy input sensors to the smooth animation that Google map and other similar application draw of the cell phone.

  • @DanHaiduc
    @DanHaiduc 4 года назад

    Google hosted this amazing talk on sensor fusion, then proceeded to cripple what could be the most useful indoor localization tool: Wi-Fi scanning.
    issuetracker.google.com/issues/112688545
    issuetracker.google.com/issues/79906367

  • @dnumdug
    @dnumdug 13 лет назад

    Glad I got an simple explanation and understanding on those things... (!!).
    Is this a IQ test or what?..

  • @AskAFreemason
    @AskAFreemason 12 лет назад

    In reality is possible to recover from a 90 degree angle on an airplane.
    It's possible. But if the software is using Euler angles you are giving zero chance for that to happen... so, please, do not use Euler angles on aviation please.

  • @NobleHiro
    @NobleHiro 7 лет назад

    Anyone want to give definitions for all the terminologies used here?

  • @DrR1pper
    @DrR1pper 11 лет назад

    lol, you bet!

  • @hallodaniel1312
    @hallodaniel1312 3 года назад

    36:55

  • @cactustactics
    @cactustactics 11 лет назад

    Yeah what a noob am I right

  • @popcat_7
    @popcat_7 3 года назад

    What the hell is a pedometer?:D 31:49

  • @IAmTheRealDrake
    @IAmTheRealDrake 14 лет назад

    2nd

  • @TheLimonerthi
    @TheLimonerthi 9 лет назад +4

    22 people love Jesus

  • @lennyhome
    @lennyhome 11 лет назад

    The demos are unimpressive and on the overall it seems to be quite useless. Good job.

  • @EiziEizz
    @EiziEizz 11 лет назад +1

    Euler is pronounced that way, because he was german you geography ignoramous.
    :)

    • @dennislange317
      @dennislange317 10 лет назад

      Germany and Switzerland isn't the same you geography ignoramous.

    • @EiziEizz
      @EiziEizz 10 лет назад

      Even though Basel is right at the border, Fair enough.
      I should have said the country of his origin spoke german, thus the pronounciation.

    • @Fogmeister
      @Fogmeister 10 лет назад

      Eizi Eizz *pronunciation :P

  • @ElTurbinado
    @ElTurbinado 11 лет назад

    I'm a little upset he didn't mention the standard magnetorometer or the speedoscope. This guy is such a noob.

  • @jianguo5228
    @jianguo5228 8 лет назад +173

    Even 6 years later, it seems to be the most comprehensive talk about sensors and sensor fusion I have ever watched. Thank you!

    • @Fatih-Ada
      @Fatih-Ada 6 лет назад +7

      Even almost 8 years later, I totally agree :)

    • @HabibRK
      @HabibRK 6 лет назад +1

      Indeed ...

    • @ArpanDasS
      @ArpanDasS 5 лет назад +2

      Even almost 9 years later...

    • @evilraccoon6538
      @evilraccoon6538 5 лет назад +2

      ok, 3 jun 2019 the same. I am happy to find this video.

    • @iyxan23
      @iyxan23 4 года назад

      Welcome to Jun 26th - 10 Years ago...

  • @TheiLame
    @TheiLame 8 лет назад

    This is 5 years old video, is it still up to date in terms of what they use? :) anyone knows?