Path planning of a mobile robot in outdoor terrain

Abstract

In this study, we discuss the path planning of mobile a robot using an aerial image. Many times mobile robots are to be deputed to go to far off lands on a mission over uneven outdoor terrains. The aerial image available either as satellite images or produced by aerial drones can be used to construct a rough path for the navigation of the mobile robot. First Gaussian Process Bayesian classifier is used to classify the different classes of terrain. Next each class is associated with a cost denoting the cost of traversal of a unit distance in that particular domain. The costs account for the energy costs, risk of accidents, etc. These numerical values corresponding to each location are called as a costmap, and that array of costmap is passed to the A* algorithm which is a graph search algorithm. The A* algorithm gives the optimal path. The final result is shown in the form of the path over the aerial image with different resolutions and samples of the image.