Examinator: Henk Wymeersch, Inst för elektroteknik
Handledare: Srikar Muppirisetty och Sohini Roychowdhury, Volvo Car Corporation samt Maryam Lashgari, Inst för elektroteknik
The simultaneous localization and mapping (SLAM) algorithms aimed for autonomous vehicles (AVs) are required to utilize sensor redundancies specific to AVs and enable accurate, fast, and repeatable estimations of pose and path trajectories. In this work, a combination of three SLAM algorithms is proposed that utilizes a different subset of available sensors such as inertial measurement unit (IMU), a gray-scale mono-camera, and a Lidar. Furthermore, a novel acceleration-based gravity direction initialization (AGI) method for the visual-inertial SLAM (VI-SLAM) algorithm is proposed. The SLAM algorithms, initialization methods for pose estimation accuracy, speed of convergence, and repeatability on the KITTI odometry sequences are analysed. The proposed VI-SLAM with AGI method achieves significant improvement in relative pose errors, i.e., less than 2% error, the convergence time is reduced to half a minute or less, and also, the convergence time variability is less than 3 seconds, which makes the proposed approach a perfect solution for the AVs.