Maze solving mobile robot based on image processing and standard nodal analysis

A methodology for solving mazes with an omnidirectional mobile robot using image processing techniques and standard nodal analysis is herein reported. The main advantage and novelty is that any irregular maze plagued of irregular obstacles is acquired from an image and depending of the dimensions of the mobile robot, a resistive grid is generated with the same dimensions of the maze. Therefore, the maze size can be scaled from 1 m × 1 m to 66 m × 66 m. This scaling range can be moved down or up, depending on the size of the robot to be used. The main contribution is that standard nodal analysis is used to formulate the system of equations and by applying Kirchhoff’s current law, the shortest path can be found within the reduced resistive network due to the elimination of those nodes and branches absorbed by irregular obstacles. To avoid singular matrices during the formulation and solution of the system of equations, an undirected graph decision algorithm is used. As a consequence, the omnidirectional robot follows the collision-free shortest path from the start-point until the end-point in the main sub-network. Three studies cases are used to validate the proposed methodology, showing its effectiveness in solving a large 40 m × 40 m maze in a CPU time equal to 139.64 s, with the mobile robot traveling a distance of 278.8 m in a time of 539.64 s. Experimental results show the usefulness of the developed methodology and it can be used in applications such as agriculture, military, intelligent traffic control and so on.