Visual inertial odometry using stereo cameras. Good starting point for visual inertial odometry.
- OpenCV
- Eigen
In the figure below, blue line is the VIO's position estimation error, red line is the prediction of that error using a simple linear regression with 12 independent variables (corresponding to camera's egomotion and features' distribution). The regression is performed with 560 frames (where we have the groundtruth) using a 80-20 train-test split. There's a strong implication that machine learning can significantly improve VIO's position estimation without adding any computational burden.
