Simultaneous Localization and Mapping of a small-scale vehicle using low-cost IMU, optical wheel encoders and LiDAR

Abstract: This thesis explores and investigates the usage of a low cost six Degrees ofFreedom Inertial Measurement Unit, low cost photo-electric wheel encodersand one of the cheapest available single-band Light Detection and Rangingsensor to estimate the orientation and position of a small scale vehicle. Furthermore,we develop a Simultaneous Localization and Mapping approach forour sensor suite. The Inertial and Measurement Unit provides orientation ofthe vehicle based on an Attitude and Heading Reference System algorithm, usingGradient Descent and Kalman Filter design. The position of the vehicle isestimated separately by the Inertial MeasurementUnit and wheel encoders, butthen fused together in an Extended Kalman Filter. The final odometry resultsare later matched with Light Detection and Ranging scans using gmapping.Gmapping is a ROS-package that utilizes a Rao-Blackwellized Particle Filterfor Simultaneous Localization and Mapping.The thesis will present the low-cost system in detail and illustrate the performanceof the system with empirical results. Each real-time experiment is comparedand evaluated with a Motion Capture System, providing sub-millimetreaccuracy, and presented with numerical and graphical results along with errorpropagation.