Special Topics - The Kalman Filter (18 of 55) What is a Covariance Matrix?

Поделиться
HTML-код
  • Опубликовано: 23 дек 2024

Комментарии • 44

  • @NguyenKhanhQuynh123
    @NguyenKhanhQuynh123 6 лет назад +7

    I would like to express my deepest gratitude to you, Mr. Biezen, for this marvelous series. Although, I have read a lot about Kalman filter, but not until watching these videos, that I could understand the Filter. Thank you so much.

  • @meshackamimo1945
    @meshackamimo1945 9 лет назад +33

    this is just marvellous! I have no words to xpress my gratitude. I am trying to use what I know so far to model a time series prediction, using a groundwater Level dataset...thus is just too good to be true.
    kalman filtering has now been demystified!
    it ceases to be a domain of aeronautical engineers n rocket scientists!
    we now know what kalman filtering is all about... wish the textbooks out there on the subject would be rid of the technical jargons.
    once again, sir, thanks n may God bless you immensely.

  • @thomasvanek26
    @thomasvanek26 9 лет назад

    First of all: AWESOME KALMAN FILTER LECTION!
    Could you please explain why multiplying P(k-1) with A and AT (transposed) or what the effect is!? And what matrix H (calculating the kalman gain) is and why multiplying with the transposed to P(k-1) or why calculating HP(k)HT.
    That would be so great!
    I have to write about it till next week.
    Thank you very much!

  • @Nuclear_Man_D
    @Nuclear_Man_D 4 года назад +2

    Your videos are so helpful, thank you so much!

  • @ernest987987
    @ernest987987 9 лет назад +6

    Amazing ! Keep up the good work ! Looking forward to your next lectures :)

  • @rubenguerrerorivera7462
    @rubenguerrerorivera7462 2 года назад +2

    Wow, pretty clear indeed! Awesome lecture!

  • @tomasortizcolchon9944
    @tomasortizcolchon9944 3 года назад +1

    Is there any form of knowing the Q noise of a bias from a measurment that I already know it should be zero,? using the mean and the covariance I have estimate the bias and the noise of the measure but I don't know if there is a way of estimating some noise power for bias because right now my Q matrix is like [ Qmeas, 0; 0 0] and I would like to see the effect of designing EKF with a process noise cov matrix like [Qmeas, 0; 0 Qbias]

  • @jessstuart7495
    @jessstuart7495 5 лет назад +1

    It wasn't made clear how the matrix A is used to calculate the covariance matrix P. Is the Kalman Gain a vector of weights or a single value? What is H??? Hopefully these will be explained in a later video.

  • @Kattlesahn
    @Kattlesahn 7 лет назад +4

    Hi professor,
    I notice that in this video you call P the State Covariance Matrix. But in subsequent videos you call it the Process Covariance Matrix. Are these names interchangeable?

  • @moonyounglee7546
    @moonyounglee7546 4 года назад

    yes these nomenclature were confusing. but you helped me get a clear intuition. on what these terms mean. thank you!!!

  • @ahmedemadeldean
    @ahmedemadeldean 7 месяцев назад +2

    excellent as usual!

  • @rynhandani669
    @rynhandani669 4 года назад +1

    is there a formula for finding the value 'Q'?

  • @vashistnarayansingh5995
    @vashistnarayansingh5995 4 года назад +1

    What is H in the kalman gain equation

  • @miliandmichaelstangeland5327
    @miliandmichaelstangeland5327 3 года назад +1

    So much is so clear... but what is H??

    • @MichelvanBiezen
      @MichelvanBiezen  3 года назад +1

      H is a matrix needed to convert P into a matrix format and make it compatible with the denominator in matrix format.

  • @mohammadabdollahzadeh268
    @mohammadabdollahzadeh268 2 года назад +1

    hi professor i have a question can you tell me how I can consider the Q matrix for a buck converter i have tried a lot of data and I tested them, but in to the plant but the matrix error is too large how I can find a proper covariance matrix (Q) for any system i want to implement Kalman for the system please help me thanks

    • @oumaymadridi8772
      @oumaymadridi8772 2 года назад

      Hello, I have the same question Mohammed ! If you have found a solution please help me ! thank you

    • @mohammadabdollahzadeh268
      @mohammadabdollahzadeh268 2 года назад

      @@oumaymadridi8772 hello I found that and it doesn’t matter because you should use the state space

  • @woejozney
    @woejozney Год назад +1

    changed the shirt! love it.

  • @ThomasHaberkorn
    @ThomasHaberkorn 6 лет назад +1

    is there a physical interpretation of Q besides that it helps to prevent from P becoming 0?

    • @iforce2d
      @iforce2d 5 лет назад

      I like to think of it as very small factors that are not accounted for in the process calculation. For example, friction in bearings, some wind resistance or buffeting, side-slip in turns, battery power sag/spikes etc. Whether this is correct is another story though :)

  • @dineshgaonkar8099
    @dineshgaonkar8099 6 лет назад

    How do we get to know there is an error in measurement or predicted values? Even if there are errors how to find them?

  • @EvilSpeculator
    @EvilSpeculator 7 лет назад +52

    Came for the Kalman filter, stayed for the bowties.

  • @Surftech09
    @Surftech09 8 лет назад

    Prof. how do you get the initial numerical values for the error in the observation R. when tracking a face in a video?
    thanks

  • @anupriyamishra9687
    @anupriyamishra9687 8 лет назад

    Sir can you please explain how to use kalman filter using statistical software like R or Eviews. I have some time series data, and have applied HP filter to it in Excel. How can i use Kalman filter for the same data?

    • @uditarpit
      @uditarpit 7 лет назад

      use KFAS package in R

  • @ranchannel8385
    @ranchannel8385 4 года назад

    Excellent-Longing for the remaining

  • @souravrakshitiitm6003
    @souravrakshitiitm6003 Год назад

    If P and R are matrices, how can K be a number? Of course K cannot be ratio of matrices as shown !!

  • @akanguven114
    @akanguven114 4 года назад +1

    thanks

  • @saswatibhattacharjee7387
    @saswatibhattacharjee7387 6 лет назад

    how to find out measurment error from real data?

  • @ahmedmahdi8580
    @ahmedmahdi8580 9 лет назад +1

    thank you

  • @byatse3013
    @byatse3013 9 лет назад +1

    very nice thank you

  • @enkhmurunbayasgalan7790
    @enkhmurunbayasgalan7790 9 лет назад +1

    If your state covariance matrix is going close to zero, doesn't it mean that your model was very accurate?

    • @MichelvanBiezen
      @MichelvanBiezen  9 лет назад +6

      +Enkhmurun Bayasgalan
      It is more an indication that your measurement errors are large and your model is therefore better at prediction the next state.

    • @parthi2929
      @parthi2929 6 лет назад

      your name very much sounds like one from my state Tamil Nadu but just with a different spelling/phonetics. "Elangumaran" is a very typical tamil name and comes close to your name "Enkhmurun".

  • @franktehtank8078
    @franktehtank8078 4 года назад

    saved my butt

  • @romitjivani4367
    @romitjivani4367 2 года назад +2

    Your videos are so helpful, thank you so much!