Rethinking real-time risk prediction from multi-step time series forecasting on highway car-following scenarios.

Driving risk prediction emerges as a pivotal technology within the driving safety domain, facilitating the formulation of targeted driving intervention strategies to enhance driving safety. The driving safety undergoes continuous evolution in response to the complexities of the traffic environment, representing a dynamic and ongoing serialization process. The evolutionary trend of this sequence offers valuable information pertinent to driving safety research. However, existing research on driving risk prediction has primarily concentrated on forecasting a single index, such as the driving safety level or the extreme value within a specified future timeframe. This approach often neglects the intrinsic properties that characterize the temporal evolution of driving safety. Leveraging the high-D natural driving dataset, this study employs the multi-step time series forecasting methodology to predict the risk evolution sequence throughout the car-following process, elucidates the benefits of the multi-step time series forecasting approach, and contrasts the predictive efficacy on driving safety levels across various temporal windows. The empirical findings demonstrate that the time series prediction model proficiently captures essential dynamics such as risk evolution trends, amplitudes, and turning points. Consequently, it provides predictions that are significantly more robust and comprehensive than those obtained from a single risk index. The TsLeNet proposed in this study integrates a 2D convolutional network architecture with a dual attention mechanism, adeptly capturing and synthesizing multiple features across time steps. This integration significantly enhances the prediction precision at each temporal interval. Comparative analyses with other mainstream models reveal that TsLeNet achieves the best performance in terms of prediction accuracy and efficiency. Concurrently, this research undertakes a comprehensive analysis of the temporal distribution of errors, the impact pattern of features on risk sequence, and the applicability of interaction features among surrounding vehicles. The adoption of multi-step time series forecasting approach not only offers a novel perspective for analyzing and exploring driving safety, but also furnishes the design and development of targeted driving intervention systems.

An Emergency Driving Intervention System Designed for Driver Disability Scenarios Based on Emergency Risk Field.

Driver disability has become an increasing factor leading to traffic accidents, especially for commercial vehicle drivers who endure high mental and physical pressure because of long periods of work. Once driver disability occurs, e.g., heart disease or heat stroke, the loss of driving control may lead to serious traffic incidents and public damage. This paper proposes a novel driving intervention system for autonomous danger avoidance under driver disability conditions, including a quantitative risk assessment module named the Emergency Safety Field (ESF) and a motion-planning module. The ESF considers three factors affecting hedging behavior: road boundaries, obstacles, and target position. In the field-based framework, each factor is modeled as an individual risk source generating repulsive or attractive force fields. Individual risk distributions are regionally weighted and merged into one unified emergency safety field denoting the level of danger to the ego vehicle. With risk evaluation, a path-velocity-coupled motion planning module was designed to generate a safe and smooth trajectory to pull the vehicle over. The results of our experiments show that the proposed algorithms have obvious advantages in success rate, efficiency, stability, and safety compared with the traditional method. Validation on multiple simulation and real-world platforms proves the feasibility and adaptivity of the module in traffic scenarios.

RCT protocol for driving performance in people with Parkinson's using autonomous in-vehicle technologies.

Introduction: Driving is an essential facilitator of independence, community participation, and quality of life. Drivers with Parkinson's Disease (PD) make more driving errors and fail on-road evaluations more than healthy controls. In-vehicle technologies may mitigate PD-related driving impairments and associated driving errors. Establishing a rigorous study protocol will increase the internal validity and the transparency of the scientific work. Methods: We present a protocol to assess the efficacy of autonomous in-vehicle technologies (Level 1) on the driving performance of drivers with PD via a randomized crossover design with random allocation. Drivers with a PD diagnosis based on established clinical criteria (N = 105), referred by neurologists, are exposed to two driving conditions (technology activated or not) on a standardized road course as they drove a 2019 Toyota Camry. The researchers collected demographic, clinical, on-road data observational and kinematic, and video data to understand several primary outcome variables, i.e., number of speeding, lane maintenance, signaling, and total driving errors. Discussion: The protocol may enhance participant adherence, decrease attrition, provide early and accurate identification of eligible participants, ensure data integrity, and improve the study flow. One limitation is that the protocol may change due to unforeseen circumstances and assumptions upon implementation. A strength is that the protocol ensures the study team executes the planned research in a systematic and consistent way.Following, adapting, and refining the protocol will enhance the scientific investigation to quantify the nuances of driving among those with PD in the era of automated in-vehicle technologies. Trial registration: ClinicalTrials.gov NCT04660500.

'Just Drive': An Employee-Based Intervention to Reduce Distracted Driving.

Distracted driving is a major danger on today's roadways. Employers play a critical role in developing distracted driving policies and promoting a culture of workplace driving safety. The purpose of this study was to evaluate the effectiveness of an in-person work-based class to reduce distracted driving in participating employees. The "Just Drive-Take Action Against Distraction" class was designed by the UC San Diego Training, Research and Education for Driving Safety (TREDS) program to increase awareness of the dangers of distracted driving and to encourage employees to be safe and responsible drivers, both on and off the job. Participants completed pre- and post-anonymous surveys and, in a subset of attendees, volunteers were contacted via email 3 months post-intervention to complete a driving-behavior survey on Surveymonkey.com. 115 classes for 6896 employees were delivered at 54 agencies in Southern California. A total of 4928 participants completed the pre- and post-survey; 2014 n = 2263 and 2015 n = 2665. The course was found useful (85%) and engaging (85.6%). For non-commercial drivers, 55.6% of participants reported an increase of 80-100% in awareness of the dangers of distracted driving, and 67.2% reported an increase of 80-100% in their motivation to change. For commercial drivers, 71.3% reported a motivation increase of 80-100%. There were significant increases in knowledge for both groups. In the three-month follow-up survey, participants identified multiple positive changes in distracted driving behavior. This 1-h employer-supported intervention demonstrated positive changes in short-term intention and medium-term behaviors.

A Pilot Study Assessing Performance and Visual Attention of Teenagers with ASD in a Novel Adaptive Driving Simulator.

Individuals with Autism Spectrum Disorder (ASD), compared to typically-developed peers, may demonstrate behaviors that are counter to safe driving. The current work examines the use of a novel simulator in two separate studies. Study 1 demonstrates statistically significant performance differences between individuals with (N = 7) and without ASD (N = 7) with regards to the number of turning-related driving errors (p < 0.01). Study 2 shows that both the performance-based feedback group (N = 9) and combined performance- and gaze-sensitive feedback group (N = 8) achieved statistically significant reductions in driving errors following training (p < 0.05). These studies are the first to present results of fine-grained measures of visual attention of drivers and an adaptive driving intervention for individuals with ASD.