Motion Consistency Loss for Monocular Visual Odometry with Attention-Based Deep Learning
This work addresses visual odometry for robotics or autonomous systems, but it is incremental as it builds on existing deep learning methods with a new loss component.
The paper tackled the problem of improving monocular visual odometry by introducing a motion consistency loss that leverages repeated motions in consecutive video clips, resulting in increased performance on the KITTI odometry benchmark.
Deep learning algorithms have driven expressive progress in many complex tasks. The loss function is a core component of deep learning techniques, guiding the learning process of neural networks. This paper contributes by introducing a consistency loss for visual odometry with deep learning-based approaches. The motion consistency loss explores repeated motions that appear in consecutive overlapped video clips. Experimental results show that our approach increased the performance of a model on the KITTI odometry benchmark.