Multi-Scale Bag-of-Features for Scalable Map Retrieval

Abstract

Retrieving a large collection of environment maps built by mapper robots is a key problem in mobile robot self-localization. The map retrieval problem is studied from the novel perspective of the multi-scale Bag-Of-Features (BOF) approach in this paper. In general, the multi-scale approach is advantageous in capturing both the global structure and the local details of a given map. BOF map retrieval is advantageous in its compact map representation as well as the efficient map retrieval using an inverted file system. The main contribution of this paper is combining the advantages of both approaches. Our approach is based on multi cue BOF as well as packing BOF, and achieves the efficiency and compactness of the map retrieval system. Experiments evaluate the effectiveness of the techniques presented using a large collection of environment maps.

Map-to-Text: Local Map Descriptor

“…A notable exception is several shape descriptors, including shape context employed in our previous studies [15], and the polestar descriptor shown above, by which size information is explicitly included in the descriptor, which makes it possible to distinguish those objects with different sizes. Among them, this study employs polestar descriptor as it was successful in its vector quantized form (i.e., bag-of-words model) in our previous study [10].…”

“…We adopt the polestar feature for our purpose because it has several desirable properties, including viewpoint invariance and rotation independence, and has proven effective as a landmark for map matching in previous studies [10]. The extraction algorithm consists of three steps ( Fig.…”

“…This quantization process consists of three steps: (1) 10 . The above definition of threshold is motivated by the following observation.…”

“…Thus far, the BoW approach has been utilized in various map-matching tasks, ranging from view image sequence maps to 3D point cloud maps [5]- [7]. Our proposed approach is also built on the BoW system in [10], in which the BoW framework is successfully applied to the retrieval of 2D occupancy maps using rotation invariant polestar descriptors.…”

M2T: Local map descriptor

“…We adopt the polestar feature for our purpose because it has several desirable properties, including viewpoint invariance and rotation independence, and has proven effective as a landmark for map matching in previous studies [10]. The extraction algorithm consists of three steps ( Fig.2): (1) First, a set of keypoints are sampled from the raw 2D scan points.…”