Search for: All records

Automation misuse and acceptance, influenced by trust, environmental conditions, and confidence, have hindered drivers from fully benefiting from partially automated vehicles. This study investigates how driver trust changes with AV reliance, differences in mental and physiological states, and continuous measures’ effectiveness. The takeover drivers reported lower trust than the non-takeover drivers in all scenarios. Nontakeover drivers’ elevated DLPFC activation aligns with trust networks and emotion regulation. The groups also differed in neural activation preand during scenarios with the takeover group showed more PFC, V2V3, and IFC engagement pre-scenario. Gaze revealed the takeover group fixated more on the AV button or dashboard, indicating readiness to take over, while non-takeover drivers focused on the rearview mirror, reflecting situational awareness. HRV responses showed higher physiological arousal in the takeover group pre-scenario. In summary, our multimodal approach reveals takeover behavior is associated with lower trust, cognitive unloading, increased stress, and anticipatory visual attention. 

Driver reliance on automated vehicles (AV) is a critical component of safety particularly during high-risk traffic scenarios. Factors that influence reliance, including trust, situation awareness, fatigue, and demographics, have been independently explored; however, few analyses have investigated predicting AV reliance and compared factors comprehensively. The goals of this study were to develop a random forest (RF) model to predict reliance and to analyze the importance of factors for reliance decisions. We leveraged data from a driving simulation study where participants encountered four traffic events including responding to an illegal vehicle crossing, managing construction zones, stopping at a vandalized stop sign, and a pedestrian detection task. The dataset included reliance decisions and subjective assessments of dispositional trust, situational trust, fatigue, and workload. An RF model fit to the dataset using cross validation achieved an average AUC of 0.81 and accuracy of 0.77 and situational trust emerged as the most influential predictor. 

Background Stress affects learning during training, and virtual reality (VR) based training systems that manipulate stress can improve retention and retrieval performance for firefighters. Brain imaging using functional Near Infrared Spectroscopy (fNIRS) can facilitate development of VR-based adaptive training systems that can continuously assess the trainee’s states of learning and cognition. Objective The aim of this study was to model the neural dynamics associated with learning and retrieval under stress in a VR-based emergency response training exercise. Methods Forty firefighters underwent an emergency shutdown training in VR and were randomly assigned to either a control or a stress group. The stress group experienced stressors including smoke, fire, and explosions during the familiarization and training phase. Both groups underwent a stress memory retrieval and no-stress memory retrieval condition. Participant’s performance scores, fNIRS-based neural activity, and functional connectivity between the prefrontal cortex (PFC) and motor regions were obtained for the training and retrieval phases. Results The performance scores indicate that the rate of learning was slower in the stress group compared to the control group, but both groups performed similarly during each retrieval condition. Compared to the control group, the stress group exhibited suppressed PFC activation. However, they showed stronger connectivity within the PFC regions during the training and between PFC and motor regions during the retrieval phases. Discussion While stress impaired performance during training, adoption of stress-adaptive neural strategies (i.e., stronger brain connectivity) were associated with comparable performance between the stress and the control groups during the retrieval phase. 

Emergency response (ER) workers perform extremely demanding physical and cognitive tasks that can result in serious injuries and loss of life. Human augmentation technologies have the potential to enhance physical and cognitive work-capacities, thereby dramatically transforming the landscape of ER work, reducing injury risk, improving ER, as well as helping attract and retain skilled ER workers. This opportunity has been significantly hindered by the lack of high-quality training for ER workers that effectively integrates innovative and intelligent augmentation solutions. Hence, new ER learning environments are needed that are adaptive, affordable, accessible, and continually available for reskilling the ER workforce as technological capabilities continue to improve. This article presents the research considerations in the design and integration of use-inspired exoskeletons and augmented reality technologies in ER processes and the identification of unique cognitive and motor learning needs of each of these technologies in context-independent and ER-relevant scenarios. We propose a human-centered artificial intelligence (AI) enabled training framework for these technologies in ER. Finally, how these human-centered training requirements for nascent technologies are integrated in an intelligent tutoring system that delivers across tiered access levels, covering the range of virtual, to mixed, to physical reality environments, is discussed. 

The nature of firefighters’ duties requires them to work for long periods under unfavorable conditions. To perform their jobs effectively, they are required to endure long hours of extensive, stressful training. Creating such training environments is very expensive and it is difficult to guarantee trainees’ safety. In this study, firefighters are trained in a virtual environment that includes virtual perturbations such as fires, alarms, and smoke. The objective of this paper is to use machine learning methods to discern encoding and retrieval states in firefighters during a visuospatial episodic memory task and explore which regions of the brain provide suitable signals to solve this classification problem. Our results show that the Random Forest algorithm could be used to distinguish between information encoding and retrieval using features extracted from fNIRS data. Our algorithm achieved an F-1 score of 0.844 and an accuracy of 79.10% if the training and testing data are obtained at similar environmental conditions. However, the algorithm’s performance dropped to an F-1 score of 0.723 and accuracy of 60.61% when evaluated on data collected under different environmental conditions than the training data. We also found that if the training and evaluation data were recorded under the same environmental conditions, the RPM, LDLPFC, RDLPFC were the most relevant brain regions under non-stressful, stressful, and a mix of stressful and non-stressful conditions, respectively.