Problem Statement
- Traffic accidents are the leading cause of non-natural death in the USA for people aged between 1-54 yrs old
- 95% of police agree that drowsy driving is a serious problem.
- Commuters waste an average of 54 hours a year stalled in traffic, study says.
- Autonomous vehicles are expensive to buy
- Non-autonomous vehicles are a majority on the roads
Giving non-autonomous vehicles some autonomous features that enhance safety without a huge capital investment is highly important to foster trust in autonomous vehicles
What it does
Our solution focuses on addressing some of the biggest concerns that drivers face during the current transition to fully autonomous vehicles
- Focuses on adding features which don’t alter vehicle
- Uses AI to create a safer driving environment
We plan to implement a semi-autonomous system involving cameras on traffic lights and on individual cars. We plan to detect how much traffic there is with the traffic lights, and detect driver drowsiness
What do we bring:
- Dynamic traffic lights: To save time, fuel and reduce congestion
- Drowsy detection software: For combating drowsy driving in non-autonomous cars without altering the car
How we built it
We used Tensorflow for ML, Keras for ML, and SUMO (Simulation of Urban MObility) for traffic simulation.
Challenges we ran into
Generating ML model. Finding datasets. Using SUMO (high learning curve for SUMO)
Accomplishments that we're proud of
Created a Traffic simulation in Python.
What we learned
- How ML works
- The importance of data
- Time Management
What's next for AI Eye
The future of this project is to implement the traffic light sensor system in every single traffic junction in the country. This will make commute times shorter.
The drowsiness detectors will be offered as an app.
The product can also make sure pedestrians are safe through signaling warnings. It can be used by law enforcement as well.
Built With
- keras
- python
- sumo
- tensorflow
Log in or sign up for Devpost to join the conversation.