Real-time embedded automotive dynamic framework using convolutional neural network.

In this paper, research on an autonomous car that can recognise its surroundings using sensors is conducted. It does not require any assistance from passengers and is guided by the advanced car system. The intelligent system makes the car self-driving, and its future technology could have ramifications in a variety of industries and situations. In the proposed scenario, the Convolutional Neural Network (CNN), TensorFlow Lite, and OpenCV are used. A fully automatic self-driving car based on level-5 self-driving technology is being considered. Correspondingly, three algorithms were used to build the autonomous vehicle: lane detection, traffic sign detection, and object detection. Lane detection detects the steering angle of road lanes, traffic signs detection detects traffic signs, and object detection detects objects using a camera image as input. Each algorithm is created a module and combined these algorithms to make the best decision possible. The module can also sense its surroundings, such as estimating the distance between the vehicle and nearby objects, navigating the vehicle, detecting light, and collecting images from the environment. The vehicle was both manual and self-driving modes. The manual mode is used to train the model. Self-driving mode is used to drive the car automatically based on the steering angle, traffic signs, and objects. The self-driving mode is used to operate the car automatically based on the instructions of the algorithms, such as steering angle, traffic signals, and recognised objects. [ABSTRACT FROM AUTHOR]