Please download the tuned datasets from their respective branches:
https://github.com/mvoellmy/vizbiz/tree/Kitti_Tuned
https://github.com/mvoellmy/vizbiz/tree/malaga_tune_v3
https://github.com/mvoellmy/vizbiz/tree/Parking_tuned_v3
https://github.com/mvoellmy/vizbiz/tree/poly-both_tuned_v2
Miniproject for building a visual odometry pipeline.
https://github.com/mvoellmy/vizbiz/blob/master/latex/vizbiz_report.pdf
https://drive.google.com/open?id=0B5IiBBfzxq2yaVhLYkdKR1NLUW8
The main.m file is located in the src folder. In order to run the visual odometry two launch procedures where implemented:
-
Debug Mode: For development and debug mode the main.m with default parameters can be executed. Numerous individual plots are displayed with insightful information about matching, inlier rejection and triangulation.
-
Simple GUI: Out of performance reasons a more compact and user-friendly display of the pipeline output was created with a GUI designed with the Matlab GUIDE application. Only the most crucial entities, like number of landmarks, are visualized for intuitive understanding.
Please follow the steps below to run the visual odometry through the GUI environment:
- adapt dataset paths in loadParameters.m
- type into MATLAB command window: gui_simple
- in Parameters panel select dataset to run on and toggle respective radio buttons
- hit Run to trigger the visual odometry
Vision Algorithms for Mobile Robotics, 2016 ETHZ.