Localization Through Particle Filter Powered Neural Network Estimated Monocular Camera Poses

TL;DR

This paper enhances monocular camera pose estimation for mobile robots by integrating particle filters with CNN outputs, significantly improving translational accuracy and trajectory smoothness despite limited depth information.

Contribution

It introduces a novel approach combining particle filters with CNN-based pose estimates to improve localization accuracy in monocular camera systems.

Findings

Translational accuracy is significantly improved.

Trajectory smoothness is enhanced.

Rotational estimates show no consistent improvement.

Abstract

The reduced cost and computational and calibration requirements of monocular cameras make them ideal positioning sensors for mobile robots, albeit at the expense of any meaningful depth measurement. Solutions proposed by some scholars to this localization problem involve fusing pose estimates from convolutional neural networks (CNNs) with pose estimates from geometric constraints on motion to generate accurate predictions of robot trajectories. However, the distribution of attitude estimation based on CNN is not uniform, resulting in certain translation problems in the prediction of robot trajectories. This paper proposes improving these CNN-based pose estimates by propagating a SE(3) uniform distribution driven by a particle filter. The particles utilize the same motion model used by the CNN, while updating their weights using CNN-based estimates. The results show that while the rotational component of pose estimation does not consistently improve relative to CNN-based estimation, the translational component is significantly more accurate. This factor combined with the superior smoothness of the filtered trajectories shows that the use of particle filters significantly improves the performance of CNN-based localization algorithms.