Intrigued by both robotics and the concept of machine learning, EyeSpy was created by four New York students who love to tinker. This tiny contraption is able to travel the land without the need for a mundane remote control. Instead, the EyeSpy uses our own custom image processing software to detect the movement of the pilot's eyes. The device tracks the movement of its pilot's pupils and determines a direction to move in. This leaves a totally hands free experience while still maintaining control of the cart!
The EyeSpy can be broken down into 3 main components; eye-movement based controls, wireless communication between the cart and console, and the mechanics of the machine itself. Using a combination of camera feeds and neural networking, the operator console generates values based on how far the driver's pupils are from center. This tells the cart whether to turn left or right on its path to greatness. The second feat we accomplished was wireless communication between our console (a laptop) and the onboard raspberry pi. We were able to achieve successful communication by using a "socket" program between the raspberry pi, which acted as a server host, and the laptop which connected to it. This two-way communication was written in python and was one of the more challenging portions of the build. Lastly, the cart is powered by two low-torque, low voltage DC motors operated by the raspberry pi. The raspberry pi is the brain of the bot and controls not only the motors but also receives directions from the console and a few onboard LEDs as well.
Through this experience, our whole team really pushed what we are capable of, and touched on all of our core strengths. In creating this hack, we were also successfully able to touch on three core fields of engineering and science; efficient mechanics, electrical circuitry, and computer coding. While the EyeSpy has its limits, the potential shown by the tech used has not gone unnoticed. We plan on furthering the development of our wireless, hands-free, vehicular control in the future, so make sure to keep an eye on the horizon.
Built With
- dc-motors
- neural-network
- opencv
- python
- raspberry-pi
- socket-(wireless-communication)
- tensorflow
Log in or sign up for Devpost to join the conversation.