Development of an algorithm for constructing the optimal path based on the floor plan for the measuring robot

Abstract

The current level of development of robotics allows the use of robots to perform many tasks. A separate class of tasks is work in conditions that pose a danger to human life and health, in particular, work at nuclear facilities with an increased background radiation. Often at such facilities there are problems with coverage of the entire area of the facility with a wireless communication signal. In such a situation, remote control circuits for the measuring robot become unavailable. One of the options for controlling the robot in such conditions is an autonomous mode of operation according to a task prepared in advance based on the plan of the object. This global task can be represented as a set of simpler subtasks, some of which will be considered in this article, in particular - digitizing an image with a floor plan and converting it into a coordinate array of obstacles; building a path from the starting point to the end point, taking into account obstacles; selection of the optimal algorithm for constructing a path based on the type of floor plan. When digitizing, the original grayscale image is initially loaded, which is then binarized using the Otsu algorithm, the resulting image is then converted into an array of obstacle coordinates. Three different algorithms are used to build the path - Dijkstra's algorithm, A* search algorithm and D* search algorithm. Using each algorithm, the path is calculated for several rooms. Next, we consider an implementation option for choosing the optimal path construction algorithm using machine learning methods to determine similar images. For each image in the pre-trained dataset, an optimal algorithm is known, which is also recommended to be used for an image that is determined to be "similar" to the corresponding image in the dataset.