Robotic mapping, localization and navigation in ROS

Abstract

Today robotics has found its way into many aspects of life, from hobbyists creating their own home made robots to industrial robots found in many factories. One particular aspect of robotics that has been becoming more popular lately is the concept of autonomously navigating robots. These are robots that are able to move around in the world on their own without a human controlling them. This dissertation details the modules that implement autonomous navigation of a robot, specifically the PowerBot research platform built by Adept MobileRobots, on the ground. This involves the analysis of algorithms for the mapping , localization and navigation in dynamic environments. The mapping procedure was implemented using Simultaneuos Localization and Mapping algorithms and results from two of the more popular ones, GMapping and Hector SLAM were compared. The localization module implemented using theAdaptive Monte Carlo Localization (AMCL) particle filter was tested in environments with multiple unique landmarks and environments with few landmarks. The results were compared to position estimates obtained from using odometry, a dead-reckoning technique using data from wheel-encoders attached to the motor shaft of the robot’s wheels. Instead of having to redevelop software code each time a robot is built, robot operating systems have been developed which allow code re-usability on different hardware. This project makes use of the open source Robot Operating System (ROS) which allows the download of software to implement the modules of the autonomous navigation process.