Design and realization of an autonomous mobile platform inspired by Segway

This thesis focuses on the control and the simultaneous localisation and mapping (SLAM) of a self-balancing Segway-type robot developed by the engineering startup Quimesis. The control of the robot and the SLAM aspect have been separated in the report for the sake of clarity. For the control part, a state of the art has been first realised concerning the regulation of the vertical position of such robot. Among the different controllers existing in the literature, the optimal type controllers retained our attention for their performance and ease of adjusting the controller parameters. Afterwards, several optimal controllers will be examined and designed based on a mathematical model of the Segway. These strategies will first be simulated in Simulink and then in the 3D Gazebo simulator with ROS. The results obtained will be compared between these two simulation environments. The goal of the control part is thus to be able to control the Segway from a joystick by sending velocity commands to the wheels. Lastly, one of the optimal regulators will be implemented and adapted to the physical robot. An analysis of the observed behavior of the real Segway will be discussed. For the SLAM part, the theoretical problem of generating a map of an unknown environment and localizing itself at the same time will first be conceptualized mathematically. Then, several algorithms used in the literature to solve this basic task in the world of robotics will be discussed through a review of the state of the art. Then, an implementation of two fully integrated packages existing in the ROS API will be performed on the Segway in the ROS-Gazebo simulator environment. These packages will be compared based on predefined metrics.