DriverSense is a tool that ensures truck driver safety by detecting drowsiness and providing real-time alerts and interventions.
What is the problem?
Drowsy driving is a significant safety concern, particularly for long-haul truck drivers who face demanding schedules and prolonged hours on the road. Fatigue impairs reaction times, decision-making, and overall driving performance, increasing the risk of accidents.
Why is this relevant?
The Goal
Our goal was to create solutions to reduce drowsy driving risks by examining factors contributing to fatigue in long-distance truck drivers. The research study aimed to analyze daily habits, including sleep, food intake, and rest breaks, and correlates them with fatigue levels.
The Solution? DriverSense
DriverSense prevents accidents from driver drowsiness through real-time monitoring and timely interventions. Using advanced optical sensors, it analyzes eye movements and blink patterns to detect fatigue. When drowsiness is detected, the system:
Activates alarms and steering wheel vibrations to alert the driver.
Suggests breaks and guides the driver to the nearest rest stop.
Notifies a designated contact in the driver’s safety network to call and support the driver.
DriverSense combines auditory alarms, vibrating steering wheels, and hands-free voice controls to ensure a safer, more attentive driving experience.
How does DriverSense Work?
Prototype Features
Optical Sensing for Drowsiness Detection
Interventions are triggered by open-source eye-tracking technology from Google AI Edge.
Haptic Steering Wheel and Alarm
Alarm sound and steering wheel vibration triggered when as first stage measure.
User has to dismiss alarm on screen through slide interaction
Alarm is used twice before sensor moves on to next phase (rest stop)
Rest Stop Suggestions
If driver is still drowsy after alarm sounds the system will ask if they would like to take the next rest stop. System provides direction to next rest stop via GPS and the system also provides information on rest stop such as food, bathrooms, etc.
Phone Call Push Notification
If driver is still drowsy the system sends a notification to a trust contact of driver. This contact is added by the driver on app: My Safety Network. There is also the option to share location to contact or send push notifications when drowsy driving is detected.
Prototype Architecture
Research Methods and Findings
Formative study
Our team began by brainstorming 100 IoT-based ideas, narrowing them down through discussion and contextual inquiries across eight areas of interest. Each team member conducted 1-2 interviews to explore specific topics, resulting in three potential problem spaces:
After deliberation, we chose to focus on truck driver fatigue. Quick 15-minute interviews with truck drivers and commuters provided insights into their schedules, fatigue, and drowsy driving behaviors.
Contextual inquiry
We found an association between heavy traffic and increased fatigue levels.
There was significant fatigue during evening drives, often exacerbated by traffic.
Truck drivers are required by regulations to take 30-minute breaks every 8 hours.
Survey
To understand and explore the factors contributing to drowsy driving among truck drivers, we conducted a diary study to analyze changes in these factors over a period of time. We recruited 4 Michigan truck drivers to document their daily activities, fatigue levels, and consumption habits over a one-week period. Data was collected through structured entries at the start of the day, during rest stops, and at the end of the day, allowing for both qualitative and quantitative analysis of behaviors and trends.
Research questions
b. Participant Quotes
c. Survey Results
Diary Study
To understand and explore the factors contributing to drowsy driving among truck drivers, we conducted a diary study to analyze changes in these factors over a period of time. We recruited 4 Michigan truck drivers to document their daily activities, fatigue levels, and consumption habits over a one-week period. Data was collected through structured entries at the start of the day, during rest stops, and at the end of the day, allowing for both qualitative and quantitative analysis of behaviors and trends.
Research Questions
Study Design
Key findings
Experience Prototype: User Enactments
User Enactment Study Design
For our user enactment study, we used insights gathered from our diary study and survey to identify four features that could potentially help mitigate challenges with drowsy driving: Alarm system, rest stop area suggestion, podcast interaction, and emergency phone calls from a close contact.
For our user enactment test, we used a speed-dating matrix to formulate potential solutions. We brainstormed product ideas, evaluated them, and merged the shortlisted concepts into a cohesive experience prototype for testing.
Key Findings
Ideal System Proposal
An ideal implementation of the Driversense Optical Sensing System would utilize open-source eye-tracking technology, such as Face Landmark Detection from Google AI Edge, to deliver a robust, real-time driver drowsiness detection solution. The system would focus on monitoring the driver’s eyelid positions (Eye Lid A and B) and detecting signs of drowsiness using a straightforward yet effective logic model.
We would like to also implement a machine-learning model that gathers drowsiness data from the driver and is able to predict when the driver is most likely to feel drowsy. This further enhances the effectiveness of the system and plays a role in creating a safer driving experience.
The haptic feedback system would be integrated through the steering wheel which would work in sync with the infotainment system of the truck. This would require minimal changes since most components of the system would be a part of the original truck.
We also plan to develop an adaptive alert mechanism that adjusts the intensity of interventions (auditory, visual, or haptic) based on the severity of the drowsiness event and the driver’s response patterns.
For example, increase vibration intensity or escalate audio alerts if the driver remains unresponsive to initial signals
Expanding the My Safety Network app into a comprehensive wellness platform for truck and long-distance drivers would create a holistic approach to driver safety and well-being. By integrating with DriverSense, the app could synchronize real-time data on drowsiness patterns and other driving behaviors, offering personalized tips and routines to help drivers stay alert and healthy.
Reflections
Limitations
Limited survey responses, with only seven participants, reduce the representativeness of the data.
The diary study was conducted for a short duration and could have been extended to seven days for deeper insights.
The system heavily relies on advanced AI for blink detection, which may face limitations in real-world conditions.
Privacy and data security risks are inherent in the collection and sharing of sensitive driver information around sleep routines and driving.
Auto-pilot systems, which could eliminate the need for driver intervention in case of drowsiness, were not explored.
Scenarios where drivers might fall asleep suddenly, without prior signs like increased blinking or drowsy eye movements, remain unaddressed.
The interventions depend heavily on family members or a support network, which might not always be feasible or reliable.
Next Steps
To improve the system, we need to gather feedback from a larger and more diverse group of drivers to ensure we’re addressing a wider range of experiences. Researching the needs of drivers with disabilities will also help us make the system more accessible. Usability testing will be a key step to refine the features and ensure everything works as intended. We also want to explore how autopilot systems could enhance safety in cases of extreme fatigue. Another important focus will be studying the types of trucks the system could be integrated into, making sure it’s practical and scalable. Lastly, we need to look into gathering data about rest stops- like parking availability, spot sizes, and amenities such as restrooms, sleep areas, and food to help provide contextual data around nearby rest stops to facilitate quick decision-making for truck drivers.
