A Stability Analysis for the Reaction Torque Observer-based Sensorless Force Control Systems

TL;DR

This paper introduces a discrete-time stability analysis for RTOb-based sensorless force control systems, highlighting the effects of parameter tuning and noise on stability and demonstrating the analysis through simulations and experiments.

Contribution

It provides a novel discrete-time stability analysis for RTOb-based force control, addressing the impact of design parameters and noise on system stability.

Findings

Bandwidth limited by velocity measurement noise and waterbed effect

Improper RTOb parameter tuning can cause non-minimum phase zeros

Stability analysis validated through simulations and experiments

Abstract

This paper proposes a new stability analysis for the Reaction Torque Observer (RTOb) based robust force control systems in the discrete-time domain. The robust force controller is implemented by employing a Disturbance Observer (DOb) to suppress disturbances, such as friction and hysteresis, in an inner-loop and another disturbance observer, viz RTOb, to estimate contact forces without using a force sensor. Since the RTOb-based robust force controllers are always implemented using computers and/or microcontrollers, this paper proposes a stability analysis in the discrete-time domain. It is shown that the bandwidth of the DOb is limited not only by the noise of velocity measurement but also by the waterbed effect. It is also shown that the stability of the robust force controller may significantly deteriorate when the design parameters of the RTOb are not properly tuned. For example, the robust force controller may have a non-minimum phase zero(s) as the design parameter of the identified inertia (torque coefficient) of the RTOb is increased (decreased). This may lead to poor stability and performance in force control applications. The proposed stability analysis conducted in the discrete-time domain is verified by simulations and experiments.