Path finding robot using image processing

Abstract

Path finding problems employ implementation of certain algorithms which enable an autonomous robot to traverse a path from a start position to an end position. Generally, the approach uses manipulation of data from the sensors to make decisions needed to traverse along the desired path. A development to this approach is the use of images of the maze to determine motion of the robot. Existing solutions either utilize supervised training or can be used only on low resolution images to ensure faster computation. Both of the approaches cannot be used to design applications to solve problems in real time. The proposed method involves image processing and an improved algorithm for path finding; which works efficiently due to its nature of acquiring the desired maze data in advance, rather than traversing the maze physically to find the path. The efficacy of the proposed method is that it is far more independent of the resolution of the captured image. Thus, it offers far more flexibility than the existing means. The proposed approach enables faster results allowing us to achieve real time computation of the possible path. Such an approach finds application in field of space exploration or even in warehouse management or home automation, where faster computation of the path required by the traversing robot.