TY - JOUR
T1 - Stable Model-based Control with Gaussian Process Regression for Robot Manipulators
AU - Beckers, Thomas
AU - Umlauft, Jonas
AU - Hirche, Sandra
N1 - Publisher Copyright:
© 2017
PY - 2017/7
Y1 - 2017/7
N2 - Computed-torque control requires a very precise dynamical model of the robot for compensating the manipulator dynamics. This allows reduction of the controller's feedback gains resulting in disturbance attenuation and other advantages. Finding precise models for manipulators is often difficult with parametric approaches, e.g. in the presence of complex friction or flexible links. Therefore, we propose a novel computed-torque control law which consists of a PD feedback and a dynamic feed forward compensation part with Gaussian Processes. For this purpose, the nonparametric Gaussian Process regression infers the difference between an estimated and the true dynamics. In contrast to other approaches, we can guarantee that the tracking error is stochastically bounded. Furthermore, if the number of training points tends to infinity, the tracking error is asymptotically stable in the large. In simulation and with an experiment, we demonstrate the applicability of the proposed control law and that it outperforms classical computed-torque approaches in terms of tracking precision.
AB - Computed-torque control requires a very precise dynamical model of the robot for compensating the manipulator dynamics. This allows reduction of the controller's feedback gains resulting in disturbance attenuation and other advantages. Finding precise models for manipulators is often difficult with parametric approaches, e.g. in the presence of complex friction or flexible links. Therefore, we propose a novel computed-torque control law which consists of a PD feedback and a dynamic feed forward compensation part with Gaussian Processes. For this purpose, the nonparametric Gaussian Process regression infers the difference between an estimated and the true dynamics. In contrast to other approaches, we can guarantee that the tracking error is stochastically bounded. Furthermore, if the number of training points tends to infinity, the tracking error is asymptotically stable in the large. In simulation and with an experiment, we demonstrate the applicability of the proposed control law and that it outperforms classical computed-torque approaches in terms of tracking precision.
KW - Adaptive system
KW - Data-based control Nonparametric methods
KW - Stability of nonlinear systems
KW - Stochastic control
KW - control
KW - robotic manipulators
UR - http://www.scopus.com/inward/record.url?scp=85031809555&partnerID=8YFLogxK
U2 - 10.1016/j.ifacol.2017.08.359
DO - 10.1016/j.ifacol.2017.08.359
M3 - Article
AN - SCOPUS:85031809555
SN - 1474-6670
VL - 50
SP - 3877
EP - 3884
JO - IFAC Proceedings Volumes (IFAC-PapersOnline)
JF - IFAC Proceedings Volumes (IFAC-PapersOnline)
IS - 1
ER -