SLAM Algorithm for Omni-Directional Robots based on ANN and EKF

This paper describes a Simultaneous Localization and Mapping (SLAM) algorithm that uses Infrared sensors, monocular camera, and motor shaft encoders to build a map of an unknown environment. The proposed algorithm is divided into three stages. First, Artificial Neural Networks (ANN) are used to analyze the sensors and camera image data to search for possible paths. Then, the camera image edges are detected using speeded up robust features (SURF) to find alternate paths. Finally, the paths from the previous two stages are compared and the best match path is found while Extended Kalman Filters (EKF) are used to estimate the robot position and orientation. The proposed algorithm is programmed using MATLAB software, interfaced with an omnidirectional robot by means of wireless communication, and validated experimentally using Robotino platform.