Self Driving Car using Android Things and Arduino

A “Prototype” of a self-driving car project using Android Things and Tensorflow.
The prototype version of the self-driving car that we have built is able to recognize the traffic signs, avoid the obstacle, and do the path planning.
We used machine learning capabilities using Tensorflow to train the model to recognize the traffic signs. The image classifier is running on Android Things. It is trained to understand traffic signs like Stop, Left, Right, as well as all the traffic lights.
We also used Arduino Mega, Arduino Uno, Google’s Android Things imx7d. We even used 6 Ultrasonic sensors for obstacle avoidance as well as gy80 for accelerometer and gyroscope. Several other sensors like LDR sensors are used to detect the light intensity and turn on the headlight based on it.
This also has a mobile app, used to control the car. The mobile app communicates to the android things using Firebase Real-time Database. We even added fingerprint authentication for locking and unlocking the car. That means you need to authenticate the fingerprint before you could control the car. In the mobile app, we have a manual mode and self-driving mode also.
Here are some photos of the “prototype” version of the self-driving car,