Image features for visual teach-and-repeat navigation in changing environments

Abstract

We present an evaluation of standard image features in the context of long-term visual teach-and-repeat navigation of mobile robots, where the environment exhibits significant changes in appearance caused by seasonal weather variations and daily illumination changes. We argue that for long-term autonomous navigation, the viewpoint-, scaleand rotation-invariance of the standard feature extractors is less important than their robustness to the mid-and longterm environment appearance changes. Therefore, we focus our evaluation on the robustness of image registration to variable lighting and naturally-occurring seasonal changes. We combine detection and description components of different image extractors and evaluate their performance on five datasets collected by mobile vehicles in three different outdoor environments over the course of one year. Moreover, we propose a trainable feature descriptor based on a combination of evolutionary algorithms and Binary Robust Independent Elementary Features, which we call GRIEF (Generated BRIEF). In terms of robustness to seasonal changes, the most promising results were achieved by the SpG/CNN and the STAR/GRIEF feature, which was slightly less robust, but faster to calculate.

Performance Evaluation of Feature Detectors and Descriptors Beyond the Visible

“…However, larger data-sets such as, [5] (covering a campus scenario over the course of 15 months) and [30] (obtained via a train) have been collected. Krajník et al [10] proposed and tested a descriptor which copes well with seasonal changes. This was evaluated on outdoor images and compared against other descriptors, over the course of a year.…”

Warped Hypertime Representations for Long-Term Autonomy of Mobile Robots

“…After one week, the robot visited the same locations every 10 minutes for one day, collecting 1152 time-stamped images used for testing. The training set images were then processed by the BRIEF method [38], which shows good robustness to appearance changes [39]. The extracted features belonging to the same locations were matched and we obtained their visibility over time, which was then processed by the temporal models evaluated.…”

VIZARD: Reliable Visual Localization for Autonomous Vehicles in Urban Outdoor Environments

“…In contrast to that, our map-tracking algorithm employs a pose prior and performs a local search in the image space. This variation in methodology leads to differing results as compared to [41]. Average number of observed landmarks (top), localization recall (middle), and translation accuracy on the NCLT datasets (left), and UP-Drive datasets (right).…”