TY - JOUR
T1 - Fuzzy-Based Approach for Contact Identification in Force-Controlled Robot Tasks. Experimental Results and Modelling Influence
AU - Oussalah, Mourad
AU - De Schutter, J
AU - Bruyninckx, H
PY - 2003/12/1
Y1 - 2003/12/1
N2 - This paper deals with the estimation of geometric parameters which characterize the contact situation during force-controlled execution of compliant robot tasks. The inputs for the estimation are measured contact force and measured velocity of the robot end-effector, each with six components. In particular, we focus on the simple vertex-face contact situation. This paper introduces a fuzzy logic approach, where the data are modelled in terms of possibility distributions, and the calculations are performed in a Kalman-filter-like manner, but with fuzzy numbers. The fuzzy computations are approximated such that the L-R fuzzy number representations are maintained during the entire iterative process. The approach provides an interval-like method where upper and lower bounds of Kalman state and covariance estimates are efficiently computed. This approach is also compared with interval Kalman filtering introduced by Chen et at. Further, a theoretical analysis of the influence of initial guess and sensor modelling is investigated. (C) 2003 Elsevier Ltd. All rights reserved.
AB - This paper deals with the estimation of geometric parameters which characterize the contact situation during force-controlled execution of compliant robot tasks. The inputs for the estimation are measured contact force and measured velocity of the robot end-effector, each with six components. In particular, we focus on the simple vertex-face contact situation. This paper introduces a fuzzy logic approach, where the data are modelled in terms of possibility distributions, and the calculations are performed in a Kalman-filter-like manner, but with fuzzy numbers. The fuzzy computations are approximated such that the L-R fuzzy number representations are maintained during the entire iterative process. The approach provides an interval-like method where upper and lower bounds of Kalman state and covariance estimates are efficiently computed. This approach is also compared with interval Kalman filtering introduced by Chen et at. Further, a theoretical analysis of the influence of initial guess and sensor modelling is investigated. (C) 2003 Elsevier Ltd. All rights reserved.
UR - http://www.scopus.com/inward/record.url?scp=0347526083&partnerID=8YFLogxK
U2 - 10.1016/j.engappai.2003.08.007
DO - 10.1016/j.engappai.2003.08.007
M3 - Article
VL - 16
SP - 691
EP - 707
JO - Engineering Applications of Artificial Intelligence
JF - Engineering Applications of Artificial Intelligence
ER -