Computer Vision Based Position and Speed Estimation for Accident Avoidance in Driverless Cars

Abstract

In the field of driverless cars, safety is the main concern. The safety systems for these cars are mainly dependent on the inputs of cameras and sensors like Light Detection and Ranging (LIDARs). Lidar is an essential component in driverless cars which creates 3D map of the surrounding and assists the car for driving. Lidar based safety systems are effective but are very expensive as the lidars are costly. Hence, there is a need to design a system which can assist the car effectively on the road, without using the lidar. This paper presents a proposed design of accident avoidance system for driverless cars, which uses computer vision techniques and works on the input of single video camera. It analyzes the video frames and shows accident warnings while on turn or lane change. The system is cost effective and can work with IoT devices having low computational power. To verify the performance of the system, we have performed experiments on various scenarios including cars at different positions moving with different speed and we obtained satisfactory results.

Computer Modeling: The Future of the Car Industry

“…Although individual models have not yet been used to predict the overall effect, the models have been correlated and are now able to be used to rate fuel consumption. Not only in the part of fuel consumption, there are still many parts of the car can be modeled by computer [2].…”