Sensor attack reconstruction for mobile robots via a switching Kalman fusion mechanism
- PDF / 1,289,344 Bytes
- 11 Pages / 547.087 x 737.008 pts Page_size
- 26 Downloads / 149 Views
ORIGINAL PAPER
Sensor attack reconstruction for mobile robots via a switching Kalman fusion mechanism Jun-Wei Zhu · Qi Wang · Wen-An Zhang · Li Yu · Xin Wang
Received: 19 April 2020 / Accepted: 18 August 2020 © Springer Nature B.V. 2020
Abstract Due to the poor security level in current industrial network, the control performance of robots may be severely affected by cyber attacks. This paper studies the sensor attack reconstruction problem of mobile robots, where a switching Kalman fusion mechanism is proposed to reconstruct the sensor attacks online. It is shown that the proposed mechanism is better than the existing extended state observer and event-triggered sensor attack reconstruction strategy. The experiment test demonstrates the effectiveness and superiority of the proposed method. Keywords Mobile robots · Sensor attack · Reconstruction · Switching Kalman fusion mechanism
1 Introduction Due to the low security level of modern industrial robots, the sensor measurements can be easily corrupted during the transmission from sensor to data processing center, which may further affect the physical process of mobile robots [1–6]. The dangers are similar J.-W. Zhu (B) · Q. Wang · W.-A. Zhang · L. Yu Department of Automation, Zhejiang University of Technology, Hangzhou 310023, China e-mail: [email protected] X. Wang School of Mathematical Science, Heilongjiang University, Harbin 150080, China e-mail: [email protected]
in the transmission of control commands or exchange process of neighbors’ information. Therefore, in recent years, the cyber-security issue for robotics has been arousing a growing interest in the control community [7–13]. To date, most of the existing works on attack reconstruction are designed for linear time-invariant (LTI) systems. The common attack reconstruction methods of LTI systems can be summarized as: the extended state observer (ESO) [14,15], the slidingmode observer [16,17], the unknown input observer [18], event-triggered projected observer [19] and so on [20–24]. For example, an integrated attack detection and reconstruction scheme based on sliding-mode observer is proposed in [25]. In [26], the sensor attack reconstruction problem is investigated for the distributed sensor networks. Besides, the switched gradient descent technique in [27] is modified to estimate sensor attacks, and the computational efficiency is also optimized. It is worth to mention that the kinematic model of mobile robot is a linear parameter-varying (LPV) system. Most of the unknown input estimation approaches for LPV systems [28,29] are extended from those originally designed for LTI systems. For instance, a K-step ESO is presented to optimize the fault estimation performance for T–S fuzzy systems in [30]. A robust ESO based on H2 /H∞ is provided in [31] for electro-rheological dampers which are modeled as a LPV system. Similarly, a proportional–integral observer is proposed in [32] for simultaneous actua-
123
J.-W. Zhu et al.
tor and sensor fault estimation problem. Authors of [33] reconstruct t
Data Loading...
