Routed roads: Probabilistic vision-based place recognition for changing conditions, split streets and varied viewpoints

Abstract

Vision-based place recognition is becoming an increasingly viable component of navigation systems for autonomous robots and personal aids. However, attaining robustness to variations in environmental conditions—such as time of day, weather and season—and camera viewpoint remains a major challenge. Featureless, sequence-based place recognition techniques have demonstrated promise, but often rely on long image sequences, manually-tuned parameters and exhaustive sequence match searching through multiple locations and image scales. In this paper, we address these deficiencies by implementing a condition-invariant, sequence-based place recognition algorithm suitable for networked environments, such as city streets, and routes with lateral platform shift, such as multiple-lane roads. We achieve this capability by augmenting the traditional 1D image database with a directed graph to describe the branching of contiguous sections of imagery at intersections. A particle filter is then used to efficiently explore these paths, as well as various lateral positions synthesized by rescaling imagery. Our proposed approach eliminates manual tuning of sequence length parameters, improves localization on branched routes, improves overall place recognition accuracy and coverage, and reduces computational requirements. We evaluated the new method against the original SeqSLAM and SMART algorithms on two day–night, road-based datasets and a summer–winter train dataset, where it attained superior precision-recall performance and coverage in all environments. Together, these contributions represent a significant step towards the provision of a robust, near parameter-free condition- and viewpoint-invariant visual place recognition capability for vehicles and robots.

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

“…SeqSLAM is demonstrated as a pure place recognition approach and does not apply loop closure to correct previous measurements even though it is possible. SeqSLAM has even been extended to integrate motion estimation between matching by using a graph structure representing the potential roads and a particle filter to maintain the consistency of the positioning [140]. This approach, SMART PF, shows better results than SeqSLAM.…”

A survey on Visual-Based Localization: On the benefit of heterogeneous data

“…Lowry et al [111,Section VII] explore exhaustively Visual SLAM methods that perform strong illumination invariance place recognition (e.g. SeqSLAM [128,152,153] or FAB-MAP [44,45,150]). Illumination perturbation are caused by three main phenomena: weather conditions and illumination changes across season, daily cycle and finally shadow casting (see figure 2a for illustration).…”

Unifying Visual Localization and Scene Recognition for People With Visual Impairment

“…To solve the cross-season visual changing of place recognition, Neubert et al [24] proposed an appearance change prediction method based on the vocabularies of superpixels. Pepperell et al [25] augmented the traditional one-dimension database with a directed graph, and used particle filter to achieve place recognition in networked environments. Abdollahyan et al [26] presented a sequence-based approach to visual localization using the partial order kernel and the pre-trained CNN (convolutional neural network) descriptor.…”