MotionHint: Self-Supervised Monocular Visual Odometry with Motion Constraints

TL;DR

MotionHint introduces a self-supervised monocular visual odometry method that incorporates a neural network-based motion model to improve accuracy by leveraging motion constraints, significantly enhancing existing systems on benchmark datasets.

Contribution

The paper proposes a novel neural network-based motion model integrated into self-supervised monocular VO to address local minima issues and improve performance.

Findings

Reduces ATE by up to 28.73% on KITTI benchmark.

Can be integrated into existing SSM-VO systems.

Significantly improves VO accuracy with motion constraints.

Abstract

We present a novel self-supervised algorithm named MotionHint for monocular visual odometry (VO) that takes motion constraints into account. A key aspect of our approach is to use an appropriate motion model that can help existing self-supervised monocular VO (SSM-VO) algorithms to overcome issues related to the local minima within their self-supervised loss functions. The motion model is expressed with a neural network named PPnet. It is trained to coarsely predict the next pose of the camera and the uncertainty of this prediction. Our self-supervised approach combines the original loss and the motion loss, which is the weighted difference between the prediction and the generated ego-motion. Taking two existing SSM-VO systems as our baseline, we evaluate our MotionHint algorithm on the standard KITTI benchmark. Experimental results show that our MotionHint algorithm can be easily applied to existing open-sourced state-of-the-art SSM-VO systems to greatly improve the performance by reducing the resulting ATE by up to 28.73%.