Design and Development of Intelligent Self-driving Car Using ROS and Machine Vision Algorithm

Abstract

The technology of self-driving cars is quite acclaimed these days. In this paper we are describing the design and development of an Intelligent Self Driving Car system. The system is capable of autonomously driving in places where there is a color difference between the road and the footpath/roadside, especially in gardens/parks. On the basis of the digital image-processing algorithm, which resulted into optimal operation of the self-driving car is based on a unique filtering and noise removal techniques implemented on the video feedback via the processing unit. We have made use of two control units, one master and other is the slave control unit in the control system. The master control unit does the video processing and filtering processes, whereas the slave control unit controls the locomotion of the car. The slave control unit is commanded by the master control unit based on the processing done on consecutive frames via Serial Peripheral Communication (SPI). Thus, via distributing operations we can achieve higher performance in comparison to having a single operational unit. The software framework management of the whole system is controlled using Robot Operating System (ROS). It is developed using ROS catkin workspace with necessary packages and nodes. The ROS was loaded on to Raspberry Pi 3 with Ubuntu Mate. The self-driving car could distinguish between the grass and the road and could maneuver on the road with high accuracy. It was able to detect frames having false sectors like shadows, and could still traverse the roads easily. Thus, self- driving cars have numerous advantages like autonomous surveillance, car- parking, accidents avoidance, less traffic congestion, efficient fuel consumption, and many more. For this purpose, our paper describes a cost-effective way for implementing self- driving cars.