An approach for Lane Keeping Assist at higher speeds using Deep Neural Network

Lane Keeping Assist (LKA) is a driver assistance feature found in autonomous vehicles that uses a camera sensor to detect the lane markings on the road and helps the vehicle to stay within the lane. An Autonomous vehicle is a vehicle that can sense its environment and navigate without human input or intervention. As we move the level of automation up the ladder ADAS feature like LKA must be able to analyze a huge amount of data in real-time and make precise and accurate control decisions. Currently, the existing feature works at lower speeds. At higher speeds this feature fails due to limited functionality of steering assistance and due to lack of better computational capability. In this paper, the proposed model uses the actual vehicle speed along with other mathematical coefficient data of the Mobileye camera sensor in the neural training with which there will be better steering decisions using a neural network. Data Generation, Training and Validation are the main three stages through which research is conducted. Each stage has several steps to follow in order to test in a real-time environment by using IPG CarMaker, SCANeR Studio and MATLAB/SIMULINK simulation software and Ego Vehicle-in-Loop (EVIL) to validate the Results.