Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation

Daniel L Boley; Erik S. Steinmetz; Karen T. Sutherland

doi:10.1007/BFb0013963

Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation

Daniel L Boley, Erik S. Steinmetz, Karen T. Sutherland

Computer Science and Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

6 Scopus citations

Abstract

In the robot navigation problem, noisy sensor data must be filtered to obtain the best estimate of the robot position. The discrete Kalman filter, commonly used for prediction and detection of signals in communication and control problems, has become a popular method to reduce the effect of uncertainty from the sensor data. However, in the domain of robot navigation, sensor readings are not only uncertain, but can also be relatively infrequent compared to traditional signal processing applications. In addition, a good initial estimate of location, critical for Kalman convergence, is often not available. Hence, there is a need for a filter that is capable of converging with a poor initial estimate and many fewer readings than the Kalman filter. To this end. we propose the use of a Recursive Total Least Squares Filter. This filter is easily updated to incorporate new sensor data, and in our experiments converged faster and to greater accuracy than the Kalman filter.

Original language	English (US)
Title of host publication	Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings
Editors	Michiel van Lambalgen, Frans Voorbraak, Leo Dorst
Publisher	Springer Verlag
Pages	221-234
Number of pages	14
ISBN (Print)	3540613765, 9783540613763
DOIs	https://doi.org/10.1007/BFb0013963
State	Published - 1996
Event	International Workshop on Reasoning with Uncertainty in Robotics, RUR 1995 - Amsterdam, Netherlands Duration: Dec 4 1995 → Dec 6 1995

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	1093
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	International Workshop on Reasoning with Uncertainty in Robotics, RUR 1995
Country/Territory	Netherlands
City	Amsterdam
Period	12/4/95 → 12/6/95

Bibliographical note

Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 1996.

Access

10.1007/BFb0013963

OpenUrl availability

Full text

Cite this

Boley, D. L., Steinmetz, E. S., & Sutherland, K. T. (1996). Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation. In M. van Lambalgen, F. Voorbraak, & L. Dorst (Eds.), Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings (pp. 221-234). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1093). Springer Verlag. https://doi.org/10.1007/BFb0013963

Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation. / Boley, Daniel L; Steinmetz, Erik S.; Sutherland, Karen T.
Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings. ed. / Michiel van Lambalgen; Frans Voorbraak; Leo Dorst. Springer Verlag, 1996. p. 221-234 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1093).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Boley, DL, Steinmetz, ES & Sutherland, KT 1996, Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation. in M van Lambalgen, F Voorbraak & L Dorst (eds), Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1093, Springer Verlag, pp. 221-234, International Workshop on Reasoning with Uncertainty in Robotics, RUR 1995, Amsterdam, Netherlands, 12/4/95. https://doi.org/10.1007/BFb0013963

Boley DL, Steinmetz ES, Sutherland KT. Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation. In van Lambalgen M, Voorbraak F, Dorst L, editors, Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings. Springer Verlag. 1996. p. 221-234. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/BFb0013963

Boley, Daniel L ; Steinmetz, Erik S. ; Sutherland, Karen T. / Recursive total least squares : An alternative to using the discrete Kalman filter in robot navigation. Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings. editor / Michiel van Lambalgen ; Frans Voorbraak ; Leo Dorst. Springer Verlag, 1996. pp. 221-234 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{437c9addef8f42d989fef44b3fee9dd5,

title = "Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation",

abstract = "In the robot navigation problem, noisy sensor data must be filtered to obtain the best estimate of the robot position. The discrete Kalman filter, commonly used for prediction and detection of signals in communication and control problems, has become a popular method to reduce the effect of uncertainty from the sensor data. However, in the domain of robot navigation, sensor readings are not only uncertain, but can also be relatively infrequent compared to traditional signal processing applications. In addition, a good initial estimate of location, critical for Kalman convergence, is often not available. Hence, there is a need for a filter that is capable of converging with a poor initial estimate and many fewer readings than the Kalman filter. To this end. we propose the use of a Recursive Total Least Squares Filter. This filter is easily updated to incorporate new sensor data, and in our experiments converged faster and to greater accuracy than the Kalman filter.",

author = "Boley, {Daniel L} and Steinmetz, {Erik S.} and Sutherland, {Karen T.}",

note = "Publisher Copyright: {\textcopyright} Springer-Verlag Berlin Heidelberg 1996.; International Workshop on Reasoning with Uncertainty in Robotics, RUR 1995 ; Conference date: 04-12-1995 Through 06-12-1995",

year = "1996",

doi = "10.1007/BFb0013963",

language = "English (US)",

isbn = "3540613765",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "221--234",

editor = "{van Lambalgen}, Michiel and Frans Voorbraak and Leo Dorst",

booktitle = "Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings",

}

TY - GEN

T1 - Recursive total least squares

T2 - International Workshop on Reasoning with Uncertainty in Robotics, RUR 1995

AU - Boley, Daniel L

AU - Steinmetz, Erik S.

AU - Sutherland, Karen T.

PY - 1996

Y1 - 1996

N2 - In the robot navigation problem, noisy sensor data must be filtered to obtain the best estimate of the robot position. The discrete Kalman filter, commonly used for prediction and detection of signals in communication and control problems, has become a popular method to reduce the effect of uncertainty from the sensor data. However, in the domain of robot navigation, sensor readings are not only uncertain, but can also be relatively infrequent compared to traditional signal processing applications. In addition, a good initial estimate of location, critical for Kalman convergence, is often not available. Hence, there is a need for a filter that is capable of converging with a poor initial estimate and many fewer readings than the Kalman filter. To this end. we propose the use of a Recursive Total Least Squares Filter. This filter is easily updated to incorporate new sensor data, and in our experiments converged faster and to greater accuracy than the Kalman filter.

AB - In the robot navigation problem, noisy sensor data must be filtered to obtain the best estimate of the robot position. The discrete Kalman filter, commonly used for prediction and detection of signals in communication and control problems, has become a popular method to reduce the effect of uncertainty from the sensor data. However, in the domain of robot navigation, sensor readings are not only uncertain, but can also be relatively infrequent compared to traditional signal processing applications. In addition, a good initial estimate of location, critical for Kalman convergence, is often not available. Hence, there is a need for a filter that is capable of converging with a poor initial estimate and many fewer readings than the Kalman filter. To this end. we propose the use of a Recursive Total Least Squares Filter. This filter is easily updated to incorporate new sensor data, and in our experiments converged faster and to greater accuracy than the Kalman filter.

UR - http://www.scopus.com/inward/record.url?scp=84880641314&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84880641314&partnerID=8YFLogxK

U2 - 10.1007/BFb0013963

DO - 10.1007/BFb0013963

M3 - Conference contribution

AN - SCOPUS:84880641314

SN - 3540613765

SN - 9783540613763

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 221

EP - 234

BT - Reasoning with Uncertainty in Robotics - International Workshop, RUR 1995, Proceedings

A2 - van Lambalgen, Michiel

A2 - Voorbraak, Frans

A2 - Dorst, Leo

PB - Springer Verlag

Y2 - 4 December 1995 through 6 December 1995

ER -

Recursive total least squares: An alternative to using the discrete Kalman filter in robot navigation

Abstract

Publication series

Other

Bibliographical note

Access

OpenUrl availability

Other files and links

Fingerprint

Cite this