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 - 1996/1/1
Y1 - 1996/1/1
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 -