Deep-Learning-Assisted Sensor with Multiple Perception Capabilities for an Intelligent Driver Assistance Monitoring System

Driver assistance systems can help drivers achieve better control of their vehicles while driving and reduce driver fatigue and errors. However, the current driver assistance devices have a complex structure and severely violate the privacy of drivers, hindering the development of driver assistance technology. To address these limitations, this article proposes an intelligent driver assistance monitoring system (IDAMS), which combines a Kresling origami structure-based triboelectric sensor (KOS-TS) and a convolutional neural network (CNN)-based data analysis. For different driving behaviors, the output signals of the KOS-TSs contain various features, such as a driver's pressing force, pressing time, and sensor triggering sequence. This study develops a multiscale CNN that employs different pooling methods to process KOS-TS data and analyze temporal information. The proposed IDAMS is verified by driver identification experiments, and the results show that the accuracy of the IDAMS in discriminating eight different users is improved from 96.25% to 99.38%. In addition, the results indicate that IDAMS can successfully monitor driving behaviors and can accurately distinguish between different driving behaviors. Finally, the proposed IDAMS has excellent hands-off detection (HOD), identification, and driving behavior monitoring capabilities and shows broad potential for application in the fields of safety warning, personalization, and human-computer interaction.

Driving Difficulties and Preferences of Advanced Driver Assistance Systems by Older Drivers With Central Vision Loss

Purpose

The purpose of this study was to investigate driving difficulties and Advanced Driver Assistance Systems (ADAS) use and preferences of drivers with and without central vision loss (CVL).

Methods

Fifty-eight drivers with CVL (71 ± 13 years) and 68 without (72 ± 8 years) completed a telephone questionnaire. They rated their perceived driving difficulty and usefulness of technology support in 15 driving situations under good (daytime) and reduced visibility conditions, and reported their use experience and preferences for 12 available ADAS technologies.

Results

Drivers with CVL reported more difficulty (P = 0.002) and greater usefulness of technology support (P = 0.003) than non-CVL drivers, especially in reduced visibility conditions. Increased driving difficulty was associated with higher perceived technology usefulness (r = 0.34, P < 0.001). Dealing with blind spot road users, glare, unexpected pedestrians, and unfamiliar areas were perceived as the most difficult tasks that would benefit from technology support. Drivers with CVL used more advanced ADAS features than non-CVL drivers (P = 0.02), preferred to own the blind spot warning, pedestrian warning, and forward collision avoidance systems, and favored ADAS support that provided both information and active intervention. The perceived benefits of and willingness to own ADAS technologies were high for both groups.

Conclusions

Drivers with CVL used more advanced ADAS and perceived greater usefulness of driver assistance technology in supporting difficult driving situations, with a strong preference for collision prevention support.

Translational Relevance

This study highlights the specific technology needs and preferences of older drivers with CVL, which can inform future ADAS development, evaluation, and training tailored to this group.

Part-Based Obstacle Detection Using a Multiple Output Neural Network

Detecting the objects surrounding a moving vehicle is essential for autonomous driving and for any kind of advanced driving assistance system; such a system can also be used for analyzing the surrounding traffic as the vehicle moves. The most popular techniques for object detection are based on image processing; in recent years, they have become increasingly focused on artificial intelligence. Systems using monocular vision are increasingly popular for driving assistance, as they do not require complex calibration and setup. The lack of three-dimensional data is compensated for by the efficient and accurate classification of the input image pixels. The detected objects are usually identified as cuboids in the 3D space, or as rectangles in the image space. Recently, instance segmentation techniques have been developed that are able to identify the freeform set of pixels that form an individual object, using complex convolutional neural networks (CNNs). This paper presents an alternative to these instance segmentation networks, combining much simpler semantic segmentation networks with light, geometrical post-processing techniques, to achieve instance segmentation results. The semantic segmentation network produces four semantic labels that identify the quarters of the individual objects: top left, top right, bottom left, and bottom right. These pixels are grouped into connected regions, based on their proximity and their position with respect to the whole object. Each quarter is used to generate a complete object hypothesis, which is then scored according to object pixel fitness. The individual homogeneous regions extracted from the labeled pixels are then assigned to the best-fitted rectangles, leading to complete and freeform identification of the pixels of individual objects. The accuracy is similar to instance segmentation-based methods but with reduced complexity in terms of trainable parameters, which leads to a reduced demand for computational resources.

Effects of penetration rates on the application of in-vehicle traffic lights at unsignalized intersections

OBJECTIVE

In-vehicle traffic lights have been demonstrated as an effective method to assist drivers in crossing unsignalized intersections, by displaying virtual traffic lights inside vehicles based on vehicle-to-vehicle communications. However, previous studies assumed that all the vehicles were equipped with vehicular communications, of which the deployment might last for decades. Therefore, it is necessary to consider the application of the system in partial deployment scenarios. This study aimed to analyze the influences of in-vehicle traffic light system on driving safety at unsignalized intersections within different penetration rates.

METHODS

Driving simulator experiments involving 18 participants were performed. Evaluation indexes including post-encroachment time, maximum acceleration stroke of ego vehicle, and times of near miss incidents were investigated for different penetration rates of vehicular communications.

RESULTS

Significant differences were observed between the conditions when no in-vehicle traffic light was used and when the penetration rate was 100% for the post-encroachment time (p = 0.02), and the maximum acceleration stroke of ego vehicle (p = 0.05). Meanwhile, it was found that the post-encroachment time slightly decreased, and the times of near miss incidents slightly increased when the penetration rate increased from 0% to 50%. However, the post-encroachment time was still longer for different penetration rates, compared to the condition when no in-vehicle traffic light was applied.

CONCLUSIONS

The results indicated that the in-vehicle traffic light system could significantly improve driving safety at unsignalized intersections when the penetration rate was 100%. For partial deployment conditions, it is essential to speed up the deployment of vehicular communication-based driver assistance systems within early deployment stage, especially when the penetration rate was within 50%, to ensure driving safety.

Sensing with Polarized LIDAR in Degraded Visibility Conditions Due to Fog and Low Clouds

LIDAR (Light Detection and Ranging) sensors are one of the leading technologies that are widely considered for autonomous navigation. However, foggy and cloudy conditions might pose a serious problem for a wide adoption of their use. Polarization is a well-known mechanism often applied to improve sensors’ performance in a dense atmosphere, but is still not commonly applied, to the best of our knowledge, in self-navigated devices. This article explores this issue, both theoretically and experimentally, and focuses on the dependence of the expected performance on the atmospheric interference type. We introduce a model which combines the well-known LIDAR equation with Stocks vectors and the Mueller matrix formulations in order to assess the magnitudes of the true target signal loss as well as the excess signal that arises from the scattering medium radiance, by considering the polarization state of the E–M (Electro-Magnetic) waves. Our analysis shows that using the polarization state may recover some of the poor performance of such systems for autonomous platforms in low visibility conditions, but it depends on the atmospheric medium type. This conclusion is supported by measurements held inside an aerosol chamber within a well-controlled and monitored artificial degraded visibility atmospheric environment. The presented analysis tool can be used for the optimization of design and trade-off analysis of LIDAR systems, which allow us to achieve the best performance for self-navigation in all weather conditions.

Lightweight Driver Monitoring System Based on Multi-Task Mobilenets

Research on driver status recognition has been actively conducted to reduce fatal crashes caused by the driver’s distraction and drowsiness. As in many other research areas, deep-learning-based algorithms are showing excellent performance for driver status recognition. However, despite decades of research in the driver status recognition area, the visual image-based driver monitoring system has not been widely used in the automobile industry. This is because the system requires high-performance processors, as well as has a hierarchical structure in which each procedure is affected by an inaccuracy from the previous procedure. To avoid using a hierarchical structure, we propose a method using Mobilenets without the functions of face detection and tracking and show this method is enabled to recognize facial behaviors that indicate the driver’s distraction. However, frames per second processed by Mobilenets with a Raspberry pi, one of the single-board computers, is not enough to recognize the driver status. To alleviate this problem, we propose a lightweight driver monitoring system using a resource sharing device in a vehicle (e.g., a driver’s mobile phone). The proposed system is based on Multi-Task Mobilenets (MT-Mobilenets), which consists of the Mobilenets’ base and multi-task classifier. The three Softmax regressions of the multi-task classifier help one Mobilenets base recognize facial behaviors related to the driver status, such as distraction, fatigue, and drowsiness. The proposed system based on MT-Mobilenets improved the accuracy of the driver status recognition with Raspberry Pi by using one additional device.