Depth Camera based Autonomous Mobile Robot for Indoor Environments

Abstract

As a consequence of the recent developments in the fields of Artificial Intelligence, Computer Vision and advancements in the processing capabilities of embedded processors, robots are gaining widespread acceptance for applications in indoor environments. The basic requirements of a robot that is meant for interior applications include the ability to localize itself in the environment and navigate through it. While localization using specific markers has its advantages, a more generic solution would involve the use of a 2D map and adequate algorithms to navigate the area using this map. This paper proposes a design for an autonomous robot that can achieve the basic requirements of navigation in a static environment using low-cost 3D cameras like Kinect. The designed robot uses particle filters to localize in a 2D map. Once localized, it uses A* algorithm for global path planning. Once a path is available, feedback from odometry is used by a PID controller to achieve motion control.