Self-Driving Car Using Image Processing and Deep Learning

Abstract

Self-driving autonomous vehicles are the solution for enhancing mobility intelligence related to driving. This project presents an effective way for implementation of a self-driving car. Proposed work is based on Artificial Intelligence, Computer Vision and Neural Networks. In our project, we are using many features such as mapping, tracking and local planning. We can successfully create a car that can demonstrate proper lane changes, parking, and U-turns on its own. The different innovations we are using are obstacles and curb detection methods, road vehicle tracker, and checking different traffic situations. This will make a robust autonomous self-driven car. It will successfully demonstrate proper parking allotment, lane changes, and automatic U-turns. We can do these using the obstacle and various curb detection method, the vehicle tracker. Self-driving cars combine a variety of sensors to perceive their surroundings, such as radar, lidar, sonar, GPS, odometry and inertial measurement units. Advanced control systems interpret sensory information to identify appropriate navigation paths, as well as obstacles and relevant signage. Long distance trucking is seen as being at the forefront of adopting and implementing the technology. We use Artificial Intelligence for recognizing and proposing the path that the autonomous car should follow for proper working. Additionally, a driverless car can reduce the time taken to reach the destination because it will take the shortest path, avoiding the traffic congestion. Human errors will be avoided thereby allowing disabled people (even blind people) to own their car.