Estimation of the sideslip angle is essential for enhancing the driving stability of an automobile while cornering. This study proposes an algorithm that estimates the sideslip angle by combining dynamics and kinematics models without requiring supplementary sensors. We employ a neural network to choose the weights for the fusion of the two models, which makes the kinematics model fully usable. Moreover, we design a new finite memory estimation algorithm to estimate the sideslip angle using the model, which is robust against modeling errors that can occur in dynamics/kinematics models. The proposed algorithm effectively addresses the deficiencies of the dynamics and kinematics models, affording accurate estimation. The performance of the proposed algorithm is evaluated by simulation using the CarSim program and experiments using data from driving tests of an actual vehicle.
8 February 2024 in IEEE Transactions on Industrial Electronics
This study proposes a novel event-triggered localization algorithm, based on finite-memory estimators (FMEs), which is robust against cyberattacks. A group of mobile robots was controlled by an operating system located at a remote site. Each robot transmitted measurement information to the operating system through wireless communication, where event-triggered transmission was adopted to save communication resources. The proposed FME-based novel event-triggering mechanism performed triggering effectively by predicting the performance of the remote localization system. Cyberattacks, such as denial of service and deception attacks, may occur and disturb the transmission of measurement information. To address this, we propose a secure localization algorithm by deriving a new FME using the Frobenius norm method, with the measurement model reflecting the effect of cyberattacks. The resulting algorithm, called secure event-triggered distributed finite-memory localization, is more robust against cyberattacks than existing algorithms. The algorithm's performance was demonstrated by comparing it with two existing event-triggered distributed localization algorithms, under the condition with/without cyberattacks, through experiments.
This paper presents a new filtering algorithm, switching extended Kalman filter bank (SEKFB), for indoor localization using wireless sensor networks. SEKFB overcomes the problem of uncertain process-noise covariance that arises when using the constant-velocity motion model for indoor localization. In the SEKFB algorithm, several extended Kalman filters (EKFs) run in parallel using a set of covariance hypotheses, and the most probable output obtained from the EKFs is selected using Mahalanobis distance evaluation. Simulations demonstrated that the SEKFB can provide accurate and reliable localization without the careful selection of process-noise covariance.
11 November 2024 in IEEE Transactions on Industrial Electronics
The movement of mobile robots can deviate from the control input due to factors such as wheel slip, drift, and unexpected motor loads, degrading the localization accuracy. To address this issue, a novel finite memory-simultaneous localization and calibration (FM-SLAC) is presented in this article. The proposed algorithm estimates any unknown component in the control input that is causing an error in the robot motion and then uses it to calibrate the localization results. The presented FM-SLAC has inherent robustness against modeling and computational errors due to its unique finite memory structure, allowing it to overcome the limitations of the existing algorithms. We conducted robot localization experiments in two scenarios where robot motion deviated from the control input to validate the performance of the proposed algorithm.