publications

2019

1. A simulation-based comparative study on lateral characteristics of trucks with double and triple trailers

   Chen, Yang, Peterson, Andrew W.,  Zhang, Ce, and Ahmadian, Mehdi

   International Journal of Vehicle Safety 2019

   Abs HTML

   This paper investigates the lateral stability and manoeuvrability in long combination vehicles (LCVs), namely semi-trucks with 28-ft doubles, 28-ft triples, and 33-ft doubles, using TruckSim. In particular, the likelihood of rollovers, rearward amplification, and off-tracking are analysed among those LCVs using the multi-domain dynamic models developed in TruckSim. The efforts to validate the truck dynamic model against test results are also included. The simulation results show that trucks with triple trailers exhibit a larger rearward amplification, higher likelihood of rollovers, and larger off-tracking than trucks with double trailers. Additionally, the results indicate that increasing the trailer length from 28 to 33 feet does not increase the likelihood of rollovers or the rearward amplification. In fact, the longer trailers provide a slight amount of additional roll stability due to their longer wheelbase.

2020

1. IEEE-EMBC

   A Computationally Efficient Multiclass Time-Frequency Common Spatial Pattern Analysis on EEG Motor Imagery

   Zhang, Ce, and Eskandarian, Azim

   In 2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC) 2020

   Abs arXiv HTML

   Common spatial pattern (CSP) is a popular feature extraction method for electroencephalogram (EEG) motor imagery (MI). This study modifies the conventional CSP algorithm to improve the multi-class MI classification accuracy and ensure the computation process is efficient. The EEG MI data is gathered from the Brain-Computer Interface (BCI) Competition IV. At first, a bandpass filter and a time-frequency analysis are performed for each experiment trial. Then, the optimal EEG signals for every experiment trials are selected based on the signal energy for CSP feature extraction. In the end, the extracted features are classified by three classifiers, linear discriminant analysis (LDA), naïve Bayes (NVB), and support vector machine (SVM), in parallel for classification accuracy comparison. The experiment results show the proposed algorithm average computation time is 37.22% less than the FBCSP (1 st winner in the BCI Competition IV) and 4.98% longer than the conventional CSP method. For the classification rate, the proposed algorithm kappa value achieved 2nd highest compared with the top 3 winners in BCI Competition IV.

2021

1. IOP-JNE

   EEG-inception: an accurate and robust end-to-end neural network for EEG-based motor imagery classification

   Zhang, Ce, Kim, Young-Keun, and Eskandarian, Azim

   Journal of Neural Engineering Mar 2021

   Abs arXiv HTML

   Objective. Classification of electroencephalography (EEG)-based motor imagery (MI) is a crucial non-invasive application in brain–computer interface (BCI) research. This paper proposes a novel convolutional neural network (CNN) architecture for accurate and robust EEG-based MI classification that outperforms the state-of-the-art methods. Approach. The proposed CNN model, namely EEG-inception, is built on the backbone of the inception-time network, which has showed to be highly efficient and accurate for time-series classification. Also, the proposed network is an end-to-end classification, as it takes the raw EEG signals as the input and does not require complex EEG signal-preprocessing. Furthermore, this paper proposes a novel data augmentation method for EEG signals to enhance the accuracy, at least by 3%, and reduce overfitting with limited BCI datasets. Main results. The proposed model outperforms all state-of-the-art methods by achieving the average accuracy of 88.4% and 88.6% on the 2008 BCI Competition IV 2a (four-classes) and 2b datasets (binary-classes), respectively. Furthermore, it takes less than 0.025 s to test a sample suitable for real-time processing. Moreover, the classification standard deviation for nine different subjects achieves the lowest value of 5.5 for the 2b dataset and 7.1 for the 2a dataset, which validates that the proposed method is highly robust. Significance. From the experiment results, it can be inferred that the EEG-inception network exhibits a strong potential as a subject-independent classifier for EEG-based MI tasks.

2. IEEE/CAA-JAS

   A Survey and Tutorial of EEG-Based Brain Monitoring for Driver State Analysis

   Zhang, Ce, and Eskandarian, Azim

   IEEE/CAA Journal of Automatica Sinica Mar 2021

   Abs arXiv HTML

   The driver’s cognitive and physiological states affect his/her ability to control the vehicle. Thus, these driver states are essential to the safety of automobiles. The design of advanced driver assistance systems (ADAS) or autonomous vehicles will depend on their ability to interact effectively with the driver. A deeper understanding of the driver state is, therefore, paramount. Electroencephalography (EEG) is proven to be one of the most effective methods for driver state monitoring and human error detection. This paper discusses EEG-based driver state detection systems and their corresponding analysis algorithms over the last three decades. First, the commonly used EEG system setup for driver state studies is introduced. Then, the EEG signal preprocessing, feature extraction, and classification algorithms for driver state detection are reviewed. Finally, EEG-based driver state monitoring research is reviewed in-depth, and its future development is discussed. It is concluded that the current EEG-based driver state monitoring algorithms are promising for safety applications. However, many improvements are still required in EEG artifact reduction, real-time processing, and between-subject classification accuracy.

3. IEEE-ITSC

   Attention-based Neural Network for Driving Environment Complexity Perception (Best Student Paper Award)

   Zhang, Ce, Eskandarian, Azim, and Du, Xuelai

   In 2021 IEEE International Intelligent Transportation Systems Conference (ITSC) Mar 2021

   Abs arXiv HTML

   The driver’s cognitive and physiological states affect his/her ability to control the vehicle. Thus, these driver states are essential to the safety of automobiles. The design of advanced driver assistance systems (ADAS) or autonomous vehicles will depend on their ability to interact effectively with the driver. A deeper understanding of the driver state is, therefore, paramount. Electroencephalography (EEG) is proven to be one of the most effective methods for driver state monitoring and human error detection. This paper discusses EEG-based driver state detection systems and their corresponding analysis algorithms over the last three decades. First, the commonly used EEG system setup for driver state studies is introduced. Then, the EEG signal preprocessing, feature extraction, and classification algorithms for driver state detection are reviewed. Finally, EEG-based driver state monitoring research is reviewed in-depth, and its future development is discussed. It is concluded that the current EEG-based driver state monitoring algorithms are promising for safety applications. However, many improvements are still required in EEG artifact reduction, real-time processing, and between-subject classification accuracy.

4. A Brain Wave-Verified Driver Alert System for Vehicle Collision Avoidance

   Riyahi, Pouria, Eskandarian, Azim, and Zhang, Ce

   SAE International Journal of Transportation Safety Apr 2021

   Abs HTML

   Collision alert and avoidance systems (CAS) could help to minimize driver errors. They are instrumental as an advanced driver-assistance system (ADAS) when the vehicle is facing potential hazards. Developing effective ADAS/CAS, which provides alerts to the driver, requires a fundamental understanding of human sensory perception and response capabilities. This research explores the premise that external stimulation can effectively improve drivers’ reaction and response capabilities. Therefore this article proposes a light-emitting diode (LED)-based driver warning system to prevent potential collisions while evaluating novel signal processing algorithms to explore the correlation between driver brain signals and external visual stimulation. When the vehicle approaches emerging obstacles or potential hazards, an LED light box flashes to warn the driver through visual stimulation to avoid the collision through braking. Thirty (30) subjects completed a driving simulator experiment under different near-collision scenarios. The Steady-State Visually Evoked Potentials (SSVEP) of the drivers’ brain signals and their collision mitigation (control performance) data were analyzed to evaluate the LED warning system’s effectiveness. The results show that (1) The proposed modified canonical correlation analysis evaluation (CCA-EVA) algorithm can detect SSVEP responses with 4.68% higher accuracy than the Adaptive Kalman filter; (2) The proposed driver monitoring and alert system produce on average a 52% improvement in time to collision (TTC), 54% improvement in reaction distance (RD), and an overall 26% reduction in collision rate as compared to similar tests without the LED warning.

5. ASME-IMECE

   A Comparative Analysis of Object Detection Algorithms in Naturalistic Driving Videos

   Zhang, Ce, and Eskandarian, Azim

   In ASME International Mechanical Engineering Congress and Exposition Nov 2021

   Abs HTML

   Intelligent vehicle research has been rapidly developing. Object detection is one of the most critical study areas for intelligent vehicle’s driving safety. This paper conducts a comparative analysis for two popular real-time object detection algorithms: Yolo-v4 and deconvolutional single shot multibox detector (DSSD) under a naturalistic driving environment. An 80-classes COCO dataset trains each neural network at first, then fine-tuned by the BDD100k dataset. The detection results are compared by True/False Positive Results, Precision-Recall Curve, and average precision @ intersection of union 50 and average precision @ intersection of union 75 results. According to the analysis results, the Yolo-v4 outperforms the DSSD algorithm in bad weather, nighttime conditions, and small object detections. The Yolo-v4 and DSSD mean average precision for the BDD dataset is 22.63 and 11.86, respectively. The Yolo-v4 precision is 90.81\% better than the DSSD algorithm, proving that the Yolo-v4 is a better fit for real-world driving environment studies.

2022

1. ArXiv

   A Quality Index Metric and Method for Online Self-Assessment of Autonomous Vehicles Sensory Perception

   Zhang, Ce, and Eskandarian, Azim

   ArXiv Nov 2022

   Abs

   Perception is critical to autonomous driving safety. Camera-based object detection is one of the most important methods for autonomous vehicle perception. Current camerabased object detection solutions for autonomous driving cannot provide feedback on the detection performance for each frame. We propose an evaluation metric, namely the perception quality index (PQI), to assess the camera-based object detection algorithm performance and provide the perception quality feedback frame by frame. The method of the PQI generation is by combining the fine-grained saliency map intensity with the object detection algorithm’s output results. Furthermore, we developed a superpixel-based attention network (SPA-NET) to predict the proposed PQI evaluation metric by using raw image pixels and superpixels as input. The proposed evaluation metric and prediction network are tested on three open-source datasets . The proposed evaluation metric can correctly assess the camera-based perception quality under the autonomous driving environment according to the experiment results. The network regression Rsquare values determine the comparison among models. It is shown that a Perception Quality Index is useful in self-evaluating a camera’s visual scene perception