Flexible model identification of the parallel robot Par2

Douat, Luiz Ricardo; Queinnec, Isabelle; García, Germain; Michelin, Micaël; Pierrot, François

doi:10.1109/iros.2010.5649161

Cited by 3 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There exist many different subspace identification algorithms: MOESP [35], IVM [38], CVA [18], N4SID [28], etc. In this work the N4SID (Numerical algorithm For Subspace IDentification) algorithm was employed because of its proven applicability to the identification of vibration system models [31], [26], [8].…”

Section: A Subspace Identification Methodsmentioning

confidence: 99%

“…According to the system representation of Figure 7, this work can be initially divided into 3 steps. The first step, briefly described above and presented in detail in [8], is the characterization of the frequency spectra of the disturbance signal d for the nominal operating point. The second step concerns the open-loop identification of the nominal model G from the piezo-actuator input voltage u to the accelerometer in the Y direction (section III).…”

Section: Function Block Representation Of the Closed-loop Systemmentioning

confidence: 99%

“…During the identification steps, a sampling rate of 1 kHz was chosen after verification that the initially adopted 2 kHz sampling rate led to stable but ill conditioned models (poles very close to each other near the real value 1 in the unit circle). According to the identification experiments performed in [1] and [8], a chirp signal was used as input excitation. It was experimentally verified that the highest main frequency modes excited in the Y direction for all considered operating points ( Figure 4) did not surpass by far the limit of 30 Hz.…”

Section: B Open-loop Piezo-actuator Identificationmentioning

confidence: 99%

“…In the current work, however, we are mainly concerned with the vibration response of the piezo-elements further used to compensate for the vibrations induced by the motion of the robot. Set as a black box connecting the piezoelectric voltage (system input) to the measured endeffector acceleration (system output), a system identification procedure is performed using state subspace techniques [20], leading to the production of a nominal model of the system [8] (section III).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Identification and Vibration Attenuation for the Parallel Robot Par2

Douat

Queinnec

García

et al. 2014

IEEE Trans. Contr. Syst. Technol.

View full text Add to dashboard Cite

Par2 is a two-degree-of-freedom parallel robot designed for high-speed and high-accuracy industrial pick-andplace operation tasks. As a result of the high acceleration trajectories, the end-effector undergoes some undesirable vibrations after reaching the stop positions, compromising its precision and leading therefore to an increase in the operation cycle time. Accelerometer sensors placed on the end-effector and piezoelectric patch actuators wrapped around the robot arms are employed in order to actively reduce these vibrations in a non-collocated closed-loop setting. After submitting the robot to an identification procedure, the obtained nominal model is used to synthesize a reduced order controller with the H ∞ Loop Shaping technique. Performance analysis as well as simulation and experimental results exhibit that vibration reduction is achieved around the nominal operating point, but fails for some extreme operating points, due to high control efforts. An antiwindup strategy is then employed to deal with the saturation of the actuator, which allows to achieve vibration attenuation on the whole operation domain, for a given configuration of the robot at the stop point.

show abstract