A Navigation and Control System for a Robot in Indoor/Outdoor Environments

Abstract

This thesis presents an approach to solving the global navigation problem of wheeled mobile robots in indoor and outdoor environments. The presented solutions are based on probabilistic approaches. In outdoor environment, the Extended Kalman Filter (EKF) is used to estimate the robot position and orientation based on wheel encoders, inertial measurement unit (IMU) and Global Positioning System (GPS) utilizing three different approaches. The three approaches are tested in a simulation environment, and one of them is verified in an experimental test. For indoor environment, where GPS signals are blocked, three different algorithms, which are based on Microsoft Kinect depth stream are proposed and tested in occupancy grid and feature-based maps both in simulation and experimental environments. First, the Particle Filter (PF) uses the raw depth data to localize the robot inside a pre-defined map. Second, EKF indoor localization based on landmarks extracted from the depth measurements, is utilized. In case the robot enters an unknown map, the third algorithm is used to estimate the robot pose as well as the landmark position based on EKF. This is known as simultaneous localization and mapping (EKF SLAM). Subsequently, an input-output state feedback linearization (I-O SFL) method is used to control the robot along the desired robot trajectory. Finally, a hybrid navigation system for indoor and outdoor environments is proposed and tested in both simulation and experimental environments. Simulation and experimental testing is performed to validate the proposed methods. It is observed that the EKF based techniques show better results than PF technique both in indoor and outdoor environments.