Improving the Accuracy of Stereo Visual Odometry Using Visual Illumination Estimation

TL;DR

This paper enhances stereo visual odometry accuracy by estimating sun position from images to reduce drift, eliminating the need for dedicated sun sensors, and demonstrating significant error reductions on urban driving data.

Contribution

It introduces a novel method to infer sun position from visual data to improve VO accuracy without additional sensors.

Findings

Up to 43% reduction in translational ARMSE.

Up to 59% reduction in final translational drift.

Effective sun estimation from images improves odometry accuracy.

Abstract

In the absence of reliable and accurate GPS, visual odometry (VO) has emerged as an effective means of estimating the egomotion of robotic vehicles. Like any dead-reckoning technique, VO suffers from unbounded accumulation of drift error over time, but this accumulation can be limited by incorporating absolute orientation information from, for example, a sun sensor. In this paper, we leverage recent work on visual outdoor illumination estimation to show that estimation error in a stereo VO pipeline can be reduced by inferring the sun position from the same image stream used to compute VO, thereby gaining the benefits of sun sensing without requiring a dedicated sun sensor or the sun to be visible to the camera. We compare sun estimation methods based on hand-crafted visual cues and Convolutional Neural Networks (CNNs) and demonstrate our approach on a combined 7.8 km of urban driving from the popular KITTI dataset, achieving up to a 43% reduction in translational average root mean squared error (ARMSE) and a 59% reduction in final translational drift error compared to pure VO alone.