Facial Detection Car

Inspiration

When I was brainstorming for hackathon ideas, Andy mentioned he made a self driving car so then I thought it would be funny to make a self-crashing car, then the idea turned into a face-crashing car.

What it does

A car with a Raspberry Pi and PiCamera that waits and detects a face causing the car to move forward at full speed.

How we built it

On the Raspberry Pi, I had to make a throttle part that took information from the camera and then produced a signal/value for the throttle to activate. Used OpenCv pretrained haarcascade file to detect faces.

Challenges we ran into

Adding code to the donkeycar was very complicated and overly difficult since they made adding code very unintuitively. Had to join the donkeycar discord but wasn't very helpful. Lots of errors like version errors, DLL errors, etc. Had to do lot of troubleshooting

Accomplishments that we're proud of

That it actually works with little problem and error.

What we learned

Got a deeper understanding of the opencv code, useful raspberry pi commands and concepts, and learned how to better troubleshoot problems.

What's next for Facial Detection Car

Implementing a better dataset so it doesn't prefer white faces, appropriate steering, use a dataset that has face masks so even people covering their faces can't run away. Adding a bluetooth speaker so that it plays an mp3 before it rams into the face.

Built With

• donkeycar
• opencv
• python
• raspberry-pi