Reset observers alleviating the peaking and the robustness tradeoffs: A case study on force estimation in teleoperation. (November 2019)

• Record Type: Journal Article
• Authors: Barreiro, Antonio; Delgado, Emma
• Is Part Of: ISA transactions. Volume 94(2019)
• Page Start: 36
• Page End: 46
• Publication Date: 2019-11
• Subjects: Reset observers -- Peaking -- Robust estimation -- Teleoperation; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2019.04.013
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: The peaking phenomenon is an undesirable effect appearing in observers and destroying controller performance. Several solutions have been proposed to mitigate peaking in state estimation. The literature shows that reset or impulsive observers are superior to linear (Luenberger) observers. However, the comparisons are based on particular choices of linear observers. This paper investigates this issue. First, comparative frameworks are proposed based on two traded-off performance indices: ensemble maximum-peak versus ensemble settling time for nominal conditions, and ensemble settling time versus size of the error asymptotic invariant set for quadratically bounded uncertain plants. Next, performance limitations of linear observers are represented by Pareto-optimal boundaries. In this way, not previously considered in the literature as far as known, the superiority of the chosen reset observer is more rigorously assessed. The framework is finally applied to force estimation in haptic teleoperation. Highlights: Peaking in linear observers: to faster convergence larger initial deviations peaks. Robustness in linear observers: to faster convergence greater size of final error invariant set. Quantitative framework for Performance limitation of all linear solutions. Overcoming the peaking and Robustness tradeoffs by using reset or impulsive observers. Apply reset observers to force estimation in haptic teleoperation under FR control.