Integral Sliding-Mode Observer-Based Disturbance Estimation for Euler–Lagrangian Systems

Abstract

In this paper, a novel integral sliding mode observer is proposed to estimate the external disturbance and velocity of Euler-Lagrangian systems. This method provides high bandwidth and precise estimation with only commanded input and position measurement. A system velocity measurement is not required to construct the sliding mode manifold. The convergence of the estimation error to zero is theoretically in finite time, which is proved by a direct Lyapunov method utilizing the passivity property of Euler-Lagrangian systems. An integral sliding manifold is designed to reduce the reaching phase, such that the robustness of the estimation is enhanced. The method has been applied to a robot manipulator to estimate the joint velocity and external contact forces in a physical human-robot task. Simulations and experiments reveal that this novel method provides fast, precise and robust estimation results and can be used to replace the measurement of an external force sensor. The successful application of this observer to a force-sensorless admittance controller for a manipulator contributes to the implementation of a sensor-free safety framework for humanrobot collaboration.

Incremental Model Predictive Control Exploiting Time-Delay Estimation for a Robot Manipulator

“…the TDE error mentioned in Remark 5, a small enough sampling period is required. In our experiments, 0.001s is used because of the setting of our experimental system, which is also commonly used in robotic systems, such as [9], [39]- [42], [45], [53]. If the sampling period increases, the tracking error increases and finally the closed-loop system becomes unstable.…”

“…1, where the order of the joint number is indicated. The mathematical model of the robot manipulator can be found in Tables II-IV in [53]. Nevertheless, note that the identified mathematical model is not used for the proposed IMPC.…”

An Online Robot Collision Detection and Identification Scheme by Supervised Learning and Bayesian Decision Theory

“…Thus, the applicability of the developed CDI scheme to different platforms needs further investigations, which motivates our future work on the adaptation of the CDI scheme to a wider range of platforms. Another interesting topic is to compensate the effects of the varying robot loads, which can be solved by dynamic-model-based disturbance estimation methods [9]. To investigate the influence of the individual uncertainties on the classification accuracy, we will also collect collision data with more experiment subjects in future work.…”

“…In the literature, a typical collision event handling pipeline usually contains two procedures, namely the precollision and the post-collision ones [4], [5]. The development of such a pipeline involves various topics, including collision avoidance [6], [7], collision force estimation [8], [9], collision detection and identification (CDI) [10], [11] and collision reaction strategy design [12], [13]. These topics are corresponding to different components in the pipeline.…”

Disturbance estimation for robotic systems using continuous integral sliding mode observer

“…Then, (10) Substituting t = 0 to (5), we have s(0) = 0 ∈ Ω s , which indicates that s(t) remains within Ω s for all time t ∈ R ≥0 . Thus, similar to the previous work, 49 the integral sliding mode variable designed in (5) eliminate the reaching phase of s(t) to Ω s . This indicates the necessity of the integral sliding mode design (5).…”