ALGORITHM FOR DETERMINING THE PATH FOR AUTONOMOUS MOVEMENT OF UNMANNED VEHICLES ALONG THE LINE

Abstract

Creating robotic systems and their programming represents a multidisciplinary field, they are subdivisions of artificial intelligence algorithms and machine learning. An increasingly popular trend is for the leaders of the automotive industry to produce vehicles with autonomous driving technology. Such vehicles are safer, their use helps to reduce the number of accidents and, accordingly, the level of injuries. The same stench is vicious as one of the solutions in the logistic tasks of 'the last mile' − folding, which is tied to the last stage of delivery. At the present stage of technology development, the development of artificial intelligence systems and the automation of such processes are relevant. The problem of autonomous vehicle control requires using specialized methods and general algorithms for machine learning, in particular, computer vision based on image processing. The article presents an algorithm for determining the path for autonomous movement of unmanned vehicles along the line. Existing methods of image processing, algorithms for their application and problems arising in image processing are considered. The presented algorithm for finding a path from a picture from a camera was tested on a computer model of a car. Such a car was prepared for the autonomous model competition in the Robotraffic competition. The task was to increase the efficiency of the existing sensors using additional equipment in the form of an external camera. The video data had to be processed programmatically to determine the path for autonomous traffic. An algorithm was created that works on the principle of sampling the entire path into parts, each of which is processed separately. Part processing is an averaging of binary information about the path width: the path segment width must be greater than the next segment. All segments in the final presentation are combined into a tree of paths from which one is selected for further movement. the algorithm can be easily modified for left or right path priority. The algorithm provides a fairly fast path finding. All calculations are carried out in real time. The advantages are simplicity and speed. The algorithm can be used to create an autonomous control system for car models. The presented algorithm for finding a path in a picture from a camera was tested on a computer model of a car. The presented algorithm can be used to create a more accurate system for controlling autonomous car models, as well as to create your own systems for helping to control real transport using a camera and mini computers. Also one of the next tasks is to develop a system for identifying road signs in the image.