Appearance-based landmark selection for efficient long-term visual localization

Abstract

In this paper, we present an online landmark selection method for distributed long-term visual localization systems in bandwidth-constrained environments. Sharing a common map for online localization provides a fleet of autonomous vehicles with the possibility to maintain and access a consistent map source, and therefore reduce redundancy while increasing efficiency. However, connectivity over a mobile network imposes strict bandwidth constraints and thus the need to minimize the amount of exchanged data. The wide range of varying appearance conditions encountered during long-term visual localization offers the potential to reduce data usage by extracting only those visual cues which are relevant at the given time. Motivated by this, we propose an unsupervised method of adaptively selecting landmarks according to how likely these landmarks are to be observable under the prevailing appearance condition. The ranking function this selection is based upon exploits landmark co-observability statistics collected in past traversals through the mapped area. Evaluation is performed over different outdoor environments, large time-scales and varying appearance conditions, including the extreme transition from day-time to night-time, demonstrating that with our appearance-dependent selection method, we can significantly reduce the amount of landmarks used for localization while maintaining or even improving the localization performance.

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

“…Note that all landmarks in the resulting multi-session map are expressed in one common frame of reference F M . Similar local localization and mapping algorithms have been used in our previous work [27], [35], to which we kindly refer the interested reader for more details.…”

Metrics for the Evaluation of localisation Robustness

“…There is a considerable body of research focusing on improving visual localisation robustness due to the brittle nature of visual features with time of the day, lighting, or other environmental variables. One example of this research is [9]. In this work, they design a ranking function and assign a quality rank for each map feature so that an adaptive feature selection policy can be enacted according to variation in the appearance of features due to events such as time of day or night.…”

View management for lifelong visual maps

“…In the second phase, the resultant subset of views (the summary map) is sent back to the navigation system and used for subse-quent SLAM runs. Bürki et al extended this approach to match loaded summary maps to lighting conditions [14], and Dymczyk et al explored linear and quadratic programming approaches for the offline computation step [15]. While these techniques do speed up computation in the vision frontend, they all require expensive offline computation.…”