| dc.description.abstract | Monocular Simultaneous Localization and Mapping (MonoSLAM), a derivative of Simultaneous Localization and Mapping (SLAM), is a navigation method for autonomous vehicles that uses only an inertial measurement unit and a camera to map the environment and localize the vehicle's position within the environment. Prior to this work, multiple different attempts have been made to generate optimal paths for SLAM, but no optimal path work has been performed specifically for MonoSLAM. The author details an optimal path generation (OPG) method designed specifically for MonoSLAM. In MonoSLAM, the vehicle gains useful data when it can detect a change in bearing to objects in the environment (also known as features). The OPG in question maximizes parallax among all visible features in the environment, with the goal of optimizing fuel usage and estimation accuracy.
In simulations comparing paths from this OPG method with typical MonoSLAM paths, it is evident that the OPG method produces extremely large fuel savings (up to 98%). These fuel savings come at the expense of estimation accuracy, however the OPG method still produces estimation performance that is acceptable for many applications. Looking forward, this work proves that it is indeed possible to improve upon the paths that are typically used in MonoSLAM. This thesis examines a two-dimensional MonoSLAM simulation only; no hardware implementation is performed. | en |
