We’re working hard on developing useful open-source robotics software. Check our GitHub for the most recent updates.

Sun-BCNN
Training and test files for a Bayesian Convolutional Neural Network (implemented in Caffe) that can infer Sun direction to within 10 degrees on most images. Preprocessed data and ground truth Sun directions for the KITTI odometry benchmarks also provided.
pykitti
This package provides a minimal set of tools for working with the KITTI dataset in Python.
Importing KITTI data is as easy as:
data = pykitti.raw('/your/dataset/dir', '2011_09_26', '0019').
MSCKF SWF Comparison
Supporting code for Lee Clement’s and Valentin Peretroukhin’s CRV paper: The Battle for Filter Supremacy: A Comparative Study of the Multi-State Constraint Kalman Filter and the Sliding Window Filter.