Abstract
In my technical project, my team develops a smart headwear system to detect and prevent driver fatigue. The primary end users for the device are long haul truck drivers, with general motorists as a secondary audience. Driver fatigue leads to several hundred fatalities a year. Furthermore, truck drivers are pushed to drive increasingly long hours, exacerbating fatigue and causing deadly accidents. Driver monitoring systems (DMSs) that combat drowsy driving have been commercially available since 2007 but are often expensive and integrated into the vehicle. My project presents an alternative, lower-cost solution in the form of a baseball cap with integrated sensors.
Current DMSs typically utilize eye-tracking to detect drowsiness. We similarly utilize an inward facing camera to detect eye closure. In addition to eye closure, we utilize heart-rate variability and head motion as additional markers of drowsiness. Literature suggests that heart rate variability increases near sleep, while jerky head motions indicate a user “nodding off”. The device is built around the Analog Devices (AD) MAX78000 microcontroller. The MAX78000 contains an onboard convolutional neural network (CNN) accelerator that allows it to perform high speed, low power AI inference. The MAX78000 runs an eye-state classification algorithm to determine when a user’s eyes are closed. We trained a pre-defined image classification model on a large dataset of eye images. Additionally, the MAX78000 processes data from an accelerometer and heart rate sensor for multimodal analysis.
We built a prototype of the cap using commercial off-the-shelf (COTS) evaluation boards. The cap successfully detected when a wearer’s eyes were closed for an extended period or when the wearer’s head jerked. We were unable to achieve functionality of the heart rate sensor. Additionally, while we designed and manufactured custom circuit boards for the MAX78000, camera, and other sensors, we could not implement any algorithms on these boards due to physical errors and firmware difficulties. Overall, we achieved promising early results and built a strong foundation for future improvements.
In my STS paper, I investigate public opinion of coal to nuclear transitions in Appalachia using a risk perception framework. As Appalachia’s coal industry declines, there is much discussion on how to best revive “coal country”. Over the past two decades, Appalachian coal production and employment have dropped as coal’s share of American energy production has sunk from over half to less than a fifth. Coal’s decline has had widespread negative effects on household finances, job prospects, and labor demographics. At the same time, nuclear power is positioned for a renaissance. While the American nuclear industry has remained stagnant for decades, a new class of advanced small modular reactors (SMRs) promises to cut upfront investment and construction times.
The nuclear industry is eyeing former coal power plant sites as potential locations for new reactors. These projects, termed coal to nuclear (C2N) transitions, intend to lower construction costs by reusing pieces of the coal plant. C2N transitions could revitalize local coal economies by bringing high-paying jobs, tax income, and other indirect economic benefits. But, simply forcing nuclear power onto a community without understanding local reception will likely lead to significant opposition.
I analyze C2N transitions from a science, technology, and society (STS) perspective. I begin by reviewing the history of nuclear power in America and assessing the economic vitality of C2N projects. I then review the history of the Appalachian coal industry to explain “coal culture” and the effects of coal’s decline. I apply cultural theory of risk and grid-group typology to analyze public perception of risk and investigate why Appalachians may support or oppose nuclear power. I find that individualist mindsets in Appalachia are likely to approve of nuclear power’s economic benefits, while a lack of trust in nuclear institutions contributes to opposition. I then propose a series of questions to reframe the argument for C2N to build public support in Appalachia.
My technical and STS projects are largely unrelated: my technical project pertains to driver safety and edge-computing innovation, while my STS paper focuses on nuclear power and risk. Despite this, the theories of risk perception I apply to my STS research could be used to investigate the reception of new driver-assistance technologies. Bringing a product to market involves not only technological success but ensuring that the market wants the product. In future work, I would apply cultural theory of risk and grid-group typology to characterize truck drivers as a market for drowsiness detection devices. By understanding dominant mindsets, I could suggest changes to encourage adoption of novel devices and minimize pushback due to perceived risks. Risk perception becomes a useful framework for analyzing the interaction between technology and the public at all scales.
