Dear Professor, A doubt in the basics. A large value in co variance matrix should imply a strong link between 2 nodes. In case of information matrix, a small value should imply a strong link right? Is my intuition about matrix inverse (thinking like it is a reciprocal) obscuring my understanding?
@CyrillStachniss
8 жыл бұрын
+Siddharthan P R Large variance = large uncertainty = small information
@siddharthanrajasekaran8977
8 жыл бұрын
+Cyrill Stachniss Thank you for your reply. I still have a small doubt. If the value of i,j th element in co-variance matrix is +1 (very high), it implies the i'th dimension has a high correlation to j'th dimension of the vector (both increase/decrease together). So why can't large covariance imply high information?
@danielebonicoli1159
5 жыл бұрын
@@siddharthanrajasekaran8977 I have a similar doubt, i understand the relationship between high variance and low information, but i don't understand what happens in terms of covariance. I mean, a normalized covariance of +1 or -1 should say to us that if we know one of the variable we perfectly know the other one, and this concept seems to resemble an high informative relationship, by contrast, if we have 0 correlation one variable doesn't tell us anything about the other, and this concept seems to fit as a lack of information
@nicolasperez4292
3 жыл бұрын
@@danielebonicoli1159 think of it this way: if some estimated position x_1 does not change depending on estimated position x_2, it basically means that x_1 is confident in its estimated position. i believe this is the correct reasoning.
@ellenamori1549
6 жыл бұрын
Prof. as per Ekf slam there is a way to calculate G_t . You also suggest that in SEIF algorithm 1st 4 line of EKF slam is copy and paste in SEIF slam. Then why there is another way to compute G_t?
@h2o11h2o
5 жыл бұрын
At 22:05, what should be filled into the matrix? A number? What if there's already a number there?
@abdelrahmanwaelhelaly1871
4 жыл бұрын
Zero. I don't get how would you have a value there
Пікірлер: 10