UC Berkeley student team dedicated to eliminating risky driving behaviors through their Distracted Driving Detection System
Without a doubt, distracted driving is the number one cause of car accidents, injuries, and deaths in the United States. Whether that is talking to passengers, texting, or reaching across the car, many drivers inadvertently put themselves and others on the road in danger. In California, 94% of accidents are caused by driver negligence1. And over 3,000 people a year in the United States are killed by distracted driving2.
While there are preventative measures, such as education, to encourage people to stay focused on the road and reduce risky behaviors, there is no auxiliary system in place to protect drivers in the case of an emergency or unpredictable situation where they may lose focus.
A student team at UC Berkeley, created through the Applied Data Science with Venture Applications: Data-X course, has created a machine learning model to help fight against these risky behaviors. Using Computer Vision and Deep Neural Networks, the team has built various models to differentiate between distracted and safe behaviors. These models have been trained to classify between the two behaviors using the State Farm Distracted Driver Detection dataset. With these results, they are able to advance the implementation of new safety features, including audio, visual, and haptic warning signals and autopilot activation, until drivers are back to a safe and focused driving position. By combining proactive and reactive safety solutions, the team hopes to reduce the chances of distracted driving.
Through this project, the team was able to develop and apply their data science skills to innovative new projects. They were able to connect with mentors from Honda and Yahoo! and work with members of the teaching staff to learn how to build convolutional neural networks and pose estimation models from scratch to develop working prototype applications. In the future, they hope to see their models integrated into vehicles and an end to distracted driving.
This semester, this team was just one of 17 groups of UC Berkeley graduate and undergraduate students participating in Data-X, a course that bridges the gap between theory and practice, by combining state-of-the-art tools, innovation processes. The course allows students to develop applied data science and use it on innovative projects, along with making connections with professionals via mentorship and connecting with global firms and organizations. The projects lean towards solving real-world problems and using innovative technology to help make the world a better place.
About the Team
Lily Sai is a senior at UC Berkeley studying Computer Science, Data Science, and Entrepreneurship & Technology.
Haochen Wu is a graduate student at UC Berkeley majoring in Transportation Engineering, with research interests in Intelligent Transportation Systems, UAVs Traffic Management and the application of RL in transportation systems.
Supattarasorn Income is a senior at UC Berkeley studying Data Science with the focus on AI & Machine Learning and Robotics.
Industry Mentors:
Nachi Nachiappan (Yahoo!)
Tony Fontana (Honda) supported by Rajeev Chhajer & Eric Bauer
Citations