Recursive Total Least Squares: An Alternative to Using the Discrete Kalman Filter in Robot Navigation
read more
Citations
Extended Target Tracking Using Polynomials With Applications to Road-Map Estimation
Robot localization from landmarks using recursive total least squares
Error-Covariance Analysis of the Total Least-Squares Problem
Depth estimation for autonomous robot navigation: A comparative approach
A Rapidly Converging Recursive Method for Mobile Robot Localization
References
Matrix Analysis
Adaptive Filter Theory
Applied Optimal Estimation
Related Papers (5)
Frequently Asked Questions (9)
Q2. What are the future works mentioned in the paper "Recursive total least squares: an alternative to using the discrete kalman filter in robot navigation" ?
Future work includes utilizing the filter in navigation problems with actual outdoor terrain data and combining its use with the higher level reasoning described in [ 16 ].
Q3. What is the main advantage of using the Kalman filter?
One of the main advantages of using the Kalman filter is that it is recursive, eliminating the necessity for storing large amounts of data.
Q4. What is the smallest eigenvalue of the covariance matrix?
some recursive TLS filters have been developed for applications in signal processing [4, 5, 20]. Davila[4] used a Kalman filter to obtain a fast update for the eigenvector corresponding to the smallest eigenvalue of the covariance matrix.
Q5. What is the main advantage of the Kalman filter?
Although originally designed as an estimator for dynamical systems, the filter is used in many applications as a static state estimator [13].
Q6. What is the smallest singular value of (A;b)?
If v = (v1; : : : ; vp)T is a right singular vector corresponding to the smallest singular value of (A;b), then it is well known that the TLS solution can be obtained by setting x = (v1; : : : ; vp 1)T=vp.
Q7. What is the solution in the least squares sense?
When error exists in both the measurement and the data matrix, the best solution in the least squares sense is often not as good as the best solution in the eigenvector sense, where the sum of the squares of the perpendicular distances from the points to the lines are minimized (Fig. 1).
Q8. Why was the RTLS filter not overdetermined?
Since there were only two measurements taken at this point, the system was not yet overdetermined, and the erroneous measures were given significant weight.
Q9. What is the average error distribution in the graphs?
The top three graphs have uniformly distributed error in of 2 and normally distributed error in t with standard deviation sd = 0, .05, and .1.