Third Eye

Inspiration

Atlanta is home to many cyclists, people who opt out of driving vehicles, and want to be guaranteed a safe commute to and from their destinations. We want to encourage more people to bike rather than scare them out of it. That is where Third Eye comes in. We are a three person team who are dedicated to ensuring the comfort of cyclists extends onto the roads of our city. We created a computer vision software that can be attached to the back of the user's bike, with wires seamlessly wrapped onto the side of the bike and an LED strip onto their handle. When a cyclist is on the road and a car is coming up from behind them, they’ll be alerted through a flash of the LED light strip. This can be incredibly useful for cases when they need to make a sharp turn or just want to know if the coast is clear to switch sides of the road.

What it does

Our program implements a camera that detects the motion of cars and alerts the cyclists by signaling LED lights to flash when motion has been detected.

How we built it

The project is made up of three main components: Raspberry Pi 3 model 8 V1.2, a raspberry pi camera module V2, and WS2812B LED Strips. We were able to implement an openCV software that detects the motion of cars (more or so rectangular objects when close). Next, we implemented software that would control the LED light strip that goes on the handle of the bike. We were able to then combine both forms of software so that when the camera is detecting motion of cars, then the LED lights will turn on to alert the cyclist. Our file is named ThirdEye.py for further details on the software embedding. Now, to get signals going from the Raspberry Pi towards the LED lights, we made a long wire made up of male and female jumper wires that will stretch across the bicycle.

Challenges we ran into

We faced challenges relating to the accuracy of the car detection algorithm, implementing it successfully, and being able to combine it with our LED algorithm. We were constantly searching for hardware parts since we were always short of something necessary. For example, HexLabs did not provide RPi cameras so we had to borrow from a friend at the very last minute and write up the code. Because of this, we feel as though the computer vision software is not as accurate as it could have been and has a few shortcomings.

Accomplishments

Eventually, we could build a sensor to alert bikers when the car is detected nearby. Moreover, it was a great opportunity to learn how to deal with hardwares even though all of the team members are computer science majors. We also successfully utilized openCV and machine learning algorithms integrated with Python and raspberry pi.

What we learned

We learned how to deal with machine learning algorithms and hardwares. Most of the team members did not have experience of utilizing openCV, so it was great for all of us to learn how it works and how to implement it in raspberry pi. And this project taught us how to work collaboratively and how to manage our time. 36 hours was short to implement both software and hardwares. However, we tried to solve this issue by communicating with teammates and breaking down what each person has to work on.