UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Enhanced stereo vision SLAM for outdoor heavy machine rotation sensing Lin, Li-Heng

Abstract

The thesis presents an enhanced stereo vision Simultaneous Localization and Mapping (SLAM) algorithm that permits reliable camera pose estimation in the presence of directional sunlight illumination causing shadows and non-uniform scene lighting. The algorithm has been developed to measure a mining rope shovel's rotation angle about its vertical axis ("swing" axis). A stereo camera is mounted externally to the shovel house (upper revolvable portion of the shovel), with a clear view of the shovel's lower carbody. As the shovel house swings, the camera revolves with the shovel house in a circular orbit, seeing differing views of the carbody top. While the shovel swings, the algorithm records observed 3D features on the carbody as landmarks, and incrementally builds a 3D map of the landmarks as the camera revolves around the carbody. At the same time, the algorithm localizes the camera with respect to this map. The estimated camera position is in turn used to calculate the shovel swing angle. The algorithm enhancements include a "Locally Maximal" Harris corner selection method which allows for more consistent feature selection in the presence of directional sunlight causing shadows and non-uniform scene lighting. Another enhancement is the use of 3D "Feature Cluster" landmarks rather than single feature landmarks, which improves the robustness of the landmark matching and reduces the SLAM filter's computational cost. The vision-based sensor's maximum swing angle error is less than +/- 1 degree upon map convergence. Results demonstrate the improvements of using the novel techniques compared to past methods.

Item Media

Item Citations and Data

Rights

Attribution-NonCommercial-NoDerivatives 4.0 International