Self-Supervised Deep Visual Stereo Odometry with 3D-Geometric Constraints
Keywords: deep learning, self-supervised learning, visual odometry, stereo image processing
Abstract: This work presents a simple but highly effective self-supervised learning framework for deep visual odometry on stereo cameras.
Recent work on deep visual odometry is often based on monocular vision.
A common approach is to use two separate neural networks, which use raw images for depth and ego-motion prediction.
This paper proposes an alternative approach that argues against separate prediction of depth and ego-motion and emphasizes the advantages of optical flow and stereo cameras.
The framework's structure is justified based on mathematical equations for image coordinate transformations.
Its central component is a deep neural network for optical flow predictions, from which depth and ego-motion can be derived.
The main contribution of this work is a 3D-geometric constraint, which enforces a realistic structure of the scene over consecutive frames and models static and moving objects.
It ensures that the neural network has to predict the optical flow as it would occur in the real world.
The presented framework is developed and tested on the KITTI dataset.
It achieves state-of-the-art results, outperforming most algorithms for deep visual odometry.
Primary Area: applications to robotics, autonomy, planning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3529
Loading