Particle Filter based SLAM to map random environments using “iRobot Roomba”

Abstract

For any mobile robot application it is important that a robot knows its location in an operating environment. The map for the operating environment may not be available every time, so the robot needs to build a map as it explores its surroundings. As a result, robot must simultaneously localize and map the operating environment. This is "Simultaneous Localization And Mapping" (SLAM) problem. SLAM finds its applications in various real life situations where automated vehicles need to map the environment during disaster relief, underwater navigation, airborne systems, minimally invasive surgery, visual tracking, etc. Statistical techniques like Kalman filters or Particle filters provide a robust framework to map an environment. Based on particle filtering, this work presents a working prototype and analysis for a SLAM implementation using an iRobot Roomba and simulations of it using MATLAB and Blender.