Development of an intelligent state machine for drone navigating from outdoor to indoor

Abstract

This report presents an intelligent state machine that enables the drone to seamlessly navigate from outdoor to indoor. Traditionally, the Global Position System (GPS) is commonly used in outdoor navigation but has a limitation in indoor or urban environments. Light Detection and Ranging (LiDAR) is widely used in indoor or urban navigation. In this project, GPS and LiDAR sensors are utilised in order for the drone to navigate smoothly from outdoor to indoor. The project focuses on developing a system to provide a pose estimate and map by fusing the Light Detection and Ranging (LiDAR), Inertial Measurement Unit (IMU) and Global Position System (GPS) sensor data. The proposed system utilised the Iterated Extended Kalman Filter (IEKF) to provide LiDAR odometry and keyframes by fusing the LiDAR and IMU measurements. The pose estimate and map are optimised by the pose graph optimisation method which uses a factor graph and incremental smoothing and mapping (iSAM2) algorithm. The proposed system is also able to perform loop detection using the Scan Context approach to correct the drifting effect. Finally, an optimised robot’s pose and map are provided by the proposed system which is able to be used for seamless outdoor to indoor transition.