CycleSense

Inspiration

During the COVID-19 pandemic, our love for biking with friends blossomed amidst the chaos. However, the joy was bittersweet as our hearing-impaired friend Robert had to brave extra dangers to explore with us. Robert enjoyed biking with us but the already risky conditions of the road was only a greater threat to him. With the goal to make life more accessible for all, this Makeathon we chose to create a prototype solution with great practical potential that will help hearing-impaired people like Robert truly experience the love for biking.

What it does

The device detects approaching vehicles from behind and uses LED lights and a motor to alert hearing-impaired cyclists. It also streams the video to a screen, offering the option for detailed viewing when needed.

How we built it

Camera Setup: We transformed a smartphone into an IP camera and established a temporary server to relay the video feed to the Raspberry Pi. Data Transmission: We employed the urllib library to fetch the video, converting it into numpy arrays for OpenCV analysis. Vehicle Detection: Utilized OpenCV algorithms analyzed the video feed to identify approaching vehicles. Tactile Alert: We used GPIO pins to integrate a vibrator, programmed to trigger upon vehicle detection. Visual Feedback: The live feed was displayed on a 5-inch LCD screen and LED array for additional visual cues.

Challenges we ran into

Hardware Issues: We faced significant setbacks with malfunctioning devices and compatibility issues. In addressing hardware issues, we were forced to simplify our setup to a single Raspberry Pi, optimizing our approach through multiple iterations. This adaptation was key in overcoming compatibility challenges and ensuring system reliability.

Communication Protocols: Establishing a reliable communication protocol for data transmission across different platforms presented its own set of challenges, requiring extensive testing and optimization.

Accomplishments that we're proud of

We're proud of the hardware-software integration system we developed, which can successfully detect vehicles while alerting the user. Our greatest pride comes from knowing this system will significantly benefit our friend Robert, enabling him to bike freely with us from now on.

What we learned

In our project, diving into OpenCV for vehicle detection was a major learning curve, teaching us the nuances of video processing. We explored Python libraries extensively to facilitate video transmission from a smartphone to the Raspberry Pi, a process that was both challenging and enlightening. Setting up a temporary server for the IP camera not only broadened our networking skills but also deepened our understanding of data communication. Incorporating tactile feedback via GPIOzero and LED array further enhanced our grasp of hardware interfaces, embodying our commitment to continuous learning and system optimization.

What's next for CycleSense

Our next steps include acquiring more powerful motors for stronger vibration alerts and refining vehicle detection with specific datasets, rather than relying on general datasets due to time constraints. We also plan to upgrade to a superior Raspberry Pi model for more stable and higher frame-rate video detection.