Made to measure: Bespoke landmarks for 24-hour, all-weather localisation with a camera

Abstract

This paper is about camera-only localisation in challenging outdoor environments, where changes in lighting, weather and season cause traditional localisation systems to fail. Conventional approaches to the localisation problem rely on point-features such as SIFT, SURF or BRIEF to associate landmark observations in the live image with landmarks stored in the map; however, these features are brittle to the severe appearance change routinely encountered in outdoor environments. In this paper, we propose an alternative to traditional point-features: we train place-specific linear SVM classifiers to recognise distinctive elements in the environment. The core contribution of this paper is an unsupervised mining algorithm which operates on a single mapping dataset to extract distinct elements from the environment for localisation. We evaluate our system on 205 km of data collected from central Oxford over a period of six months in bright sun, night, rain, snow and at all times of the day. Our experiment consists of a comprehensive N-vs-N analysis on 22 laps of the approximately 10 km route in central Oxford. With our proposed system, the portion of the route where localisation fails is reduced by a factor of 6, from 33.3% to 5.5%.

Simultaneous Localization and Mapping: A Survey of Current Trends in Autonomous Driving

“…In [164], one SVM per feature is trained across several images. The robustness is ensured by discarding the detectors not accurate or unique enough across the neighboring images.…”

Metrics for the Evaluation of localisation Robustness

“…The value varies with the parameterisation or weighting of arXiv:1904.08585v1 [cs.RO] 18 Apr 2019 the evaluation function, and its usage is restricted to visual landmark based localisation. [6] claim to have designed a visual localisation system that is can be operated at any time of the day or night, and in all weather conditions. The robustness metric they used is the portion of localisation failure in each specific dataset.…”

Efficient Indoor Positioning with Visual Experiences via Lifelong Learning

“…For instance, [24] shows that images captured at one city can be effectively distinguished from those captured at another by learning place-specific SVM classifiers on image patches. The work in [25] and [26] extends this idea to localise robots under extreme scene changes. At the training phase, for images captured from a known location, it learns the image regions that are robust to lighting and weather changes.…”