Fast, Robust, Continuous Monocular Egomotion Computation

TL;DR

This paper introduces a robust egomotion estimation method for monocular cameras that effectively handles noisy data and outliers, outperforming existing approaches on real-world datasets with minimal additional computational cost.

Contribution

The paper presents the expected residual likelihood (ERL) method for confidence weighting and a lifted kernel formulation, improving robustness and accuracy in monocular egomotion estimation.

Findings

ERL effectively identifies outliers and estimates confidence weights.

ERL outperforms lifted kernel and baseline methods on KITTI dataset.

The proposed pipeline avoids local minima with minimal runtime increase.

Abstract

We propose robust methods for estimating camera egomotion in noisy, real-world monocular image sequences in the general case of unknown observer rotation and translation with two views and a small baseline. This is a difficult problem because of the nonconvex cost function of the perspective camera motion equation and because of non-Gaussian noise arising from noisy optical flow estimates and scene non-rigidity. To address this problem, we introduce the expected residual likelihood method (ERL), which estimates confidence weights for noisy optical flow data using likelihood distributions of the residuals of the flow field under a range of counterfactual model parameters. We show that ERL is effective at identifying outliers and recovering appropriate confidence weights in many settings. We compare ERL to a novel formulation of the perspective camera motion equation using a lifted kernel, a recently proposed optimization framework for joint parameter and confidence weight estimation with good empirical properties. We incorporate these strategies into a motion estimation pipeline that avoids falling into local minima. We find that ERL outperforms the lifted kernel method and baseline monocular egomotion estimation strategies on the challenging KITTI dataset, while adding almost no runtime cost over baseline egomotion methods.