Street View Object Detection for Autonomous Car Steering Angle Prediction Using Convolutional Neural Network

Abstract

Autonomous car research is currently developing rapidly to find optimal and accurate steering angle and speed control. Various sensors such as cameras, LIDAR, and RADAR are used to recognize the surrounding environment to determine the correct steering angle prediction in avoiding obstacles. In addition to being expensive, LIDAR and RADAR have several drawbacks such as the level of accuracy that depends on the weather and the ability to detect adjacent objects. This paper will propose the prediction of steering angle and speed control in autonomous cars based on the detection of street view objects such as cars in front, traffic signs, pedestrians, and lane lines. The process of object detection and prediction of steering angle, as well as prediction of speed control using a convolutional neural network (CNN) through video captured using a single camera. In this method, other sensors such as LIDAR and RADAR are no longer needed so that the costs required are lower and the weaknesses found in LIDAR and RADAR can be eliminated. The results obtained are very good with 92% accuracy for steering angle prediction and 85% for speed control prediction. The autonomous car can run well in the simulator environment through the video taken on the real road.