Abstract
Driver fatigue and distraction are persistent contributors to serious traffic accidents, particularly among long-haul truck drivers, where crashes can result in severe injury, loss of life, and significant economic damage. Investigations by the National Transportation Safety Board (NTSB) consistently identify fatigue and delayed driver response as critical contributing factors in major incidents. Despite this awareness, there remains a lack of proactive, real-time fatigue detection systems that can reliably identify early indicators of drowsiness while remaining comfortable, unobtrusive, and acceptable for long-term use by professional drivers.
To address this gap, this capstone project developed and evaluated a wearable smart cap integrating three sensing modalities: eye movement, head motion, and heart rate. These sensors are embedded within a single non-intrusive platform and processed locally using an edge-based embedded system. A multimodal convolutional neural network analyzes synchronized sensor data in real time to classify driver fatigue. When fatigue or distraction is detected, the system provides immediate haptic feedback to alert the driver. Experimental testing demonstrated an overall fatigue-detection accuracy of 88 percent, showing that multimodal sensing improves reliability compared to single-sensor approaches and advances existing fatigue-monitoring technologies that are often reactive or intrusive.
Beyond technical performance, fatigue-monitoring technologies sit at the intersection of safety, labor management, and workplace surveillance, making their human and social interactions critical. Drivers may view monitoring devices as invasive, while fleet operators and regulators often prioritize compliance and liability reduction. These competing perspectives influence trust, adoption, and resistance, limiting the real-world effectiveness of technically capable systems.
To guide this analysis, the Social Construction of Technology framework was used to perform a case-study analysis of the Tracy Morgan crash. Through this framework, fatigue-monitoring technology is understood as something shaped by the social groups surrounding it, not just by its technical performance. Drivers, trucking companies, regulators, engineers, safety advocates, and the public each define the problem of fatigue differently. For engineers, the focus may be accurate detection of drowsiness. For drivers, the concern may be privacy, comfort, and whether monitoring becomes another form of surveillance. For companies, the issue may involve liability, cost, compliance, and productivity. For regulators and the public, the emphasis may be on enforceable standards and improved roadway safety.
Using investigation reports, regulatory documents, and secondary literature, this analysis examines how organizational culture, power dynamics, and assumptions about driver responsibility shape technology adoption. The STS research found that fatigue-monitoring technologies struggle to reach closure because the groups involved do not define the problem of fatigue in the same way. The Tracy Morgan crash shows that fatigue is often treated as an issue of driver responsibility and regulatory compliance, even though it is also shaped by company scheduling practices, economic pressure, workplace culture, and limits in existing monitoring systems. As a result, technically capable fatigue-detection systems may still face resistance if they do not address driver trust, privacy, comfort, and how the collected data will be used. When considered alongside the capstone project, this finding suggests that effective fatigue-monitoring systems must combine accurate multimodal detection with design choices that support trust, transparency, comfort, and fair use in the trucking industry.
