Steering measurement decomposition for vehicle lane keeping – A study of driver behaviour. (June 2018)

• Record Type: Journal Article
• Authors: Zhang, Yu; Gordon, Timothy; Martinez-Garcia, Miguel; Bingham, Chris
• Is Part Of: Measurement. Volume 121(2018)
• Page Start: 26
• Page End: 38
• Publication Date: 2018-06
• Subjects: PCM Pulse Control Model -- NDD Naturalistic Driving Data -- ISC Integrated Sine Component -- SC Sine Component -- EMD Empirical Mode Decomposition -- IMF Intrinsic Mode Function -- PCA Principal Component Analysis -- PC Principal Component -- SSA Singular Spectral Analysis -- RS Reconstructed Subseries -- FCF Fourier Curve-Fitting -- YRE Yaw Rate Error -- YAE Yaw Angle Error -- SVD Singular Value Decomposition -- CLK Central Lane Keeping; Pulse control model -- Empirical model decomposition -- Principal component analysis -- Singular spectral analysis -- Fourier curve-fitting -- Statistical pattern analysis; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2018.02.053
• Languages: English
• ISSNs: 0263-2241

Abstract

Graphical abstract: Highlights: The PCM can describe intermittent pulse-like qualities in real-world steering measurements. Steering angle measurements can be decomposed into a combination of trend, ISCs and SCs. EMD and PCA present appropriate tools for measurement trend extraction. SSA and FCF are appropriate techniques for the extraction of ISC and SC pulses. Patterns of pulse characteristics are useful for PCM validation and driver identification. Abstract: Steering control for vehicle lane keeping has attracted significant attention from both automotive industries and researchers. To describe intermittent pulse-like qualities imparted by drivers that are seen in real-world steering measurements, a pulse control model (PCM) is presented for vehicle lane keeping. Inspired by the PCM, a steering angle measurement is decomposed into a combination of trend, integrated sine components (ISCs) and sine components (SCs), where trend corresponds to the path curvature, ISCs to the heading angles, and SCs to the lateral positions. Trends are extracted through the use of empirical mode decomposition (EMD) and principal component analysis (PCA), with singular spectral analysis (SSA) and Fourier curve-fitting (FCF) being employed to determine the ISCs and SCs in the main pulses. Through statistical pattern analysis on experimental measurements of drivers' steering performance, it is revealed that (1) the pulse steering behaviour from real drivers shows the benefit of the proposed PCM for steering control during lane keeping, and (2) classification of pulse steering characteristics can be used for normal driver state identification and highlight abnormal driving behaviour, leading to the prospect of identifying driving characteristics typical of impaired concentration, substance misuse or tiredness, for instance.