Overview / Usage

Modern automobiles contain hundreds of sensors measuring everything from fuel level to the engine rotation, which make cars not only mechanical tools but also a smart platform built on sophisticated sensing and computing systems which generates a lot of important data. These data can be easily accessed from the CAN bus of the car.
These sensors are often used to know the current state of the car and help the car adapt to the environment. However, it is also possible to use sensors to learn about the drivers themselves. Using the generated sensor data it is possible to understand different driving styles, discover driving patterns of each individual and detect bad driving styles.

Methodology / Approach

The collected data from the CAN bus will be preprocessed and important features will be identified. Since deep learning is excelled in automatic feature extraction, it will be applied to extract important features and identify the driving pattern (just like a fingerprint) of each driver and predict who is driving the car. After finishing the identification stage the project will be extended to monitoring driving pattern and detect bad driving syles.

Technologies Used

• CAN bus data collection using a mobile app
• Deep learning and other machine learning algorithms