A pure vision-based topological SLAM system

Abstract

In this paper, we present a feasible solution to the problem of autonomous navigation in initially unknown environments using a pure vision-based approach. The mobile robot performs range sensing with a unique omnidirectional stereovision system, estimates its motion using visual odometry and detects loop closures via a place recognition system as it performs topological map building and localization concurrently. Owing to the importance of performing loop closing regularly, the mobile robot is equipped with an active loop closure detection and validation system that assists it to return to target loop closing locations, validates ambiguous loop closures and provides it with the ability to overturn the decision of an incorrectly committed loop closure. A refined incremental probabilistic framework for an appearance-based place recognition system is fully described and the final system is validated in multiple experiments conducted in indoor, semioutdoor and outdoor environments. Lastly, the performance of the probabilistic framework is compared with the rank-based framework with additional experiments conducted in the semi-autonomous mode, where the mobile robot, provided with a priori information in the form of a topological map that is built in a separate occasion in an offline manner, is required to reach its target destination.

Omnidirectional visual SLAM under severe occlusions

“…Their main contribution was to simplify the matching between features in two consecutive images through the restriction of their poses. Finally, the proposal of Lui and Jarvis [23] relies on a stereo omnidirectional (catadioptric) camera with variable baseline, so it is possible to calculate the 3D position of a feature with only one acquisition. They obtained both a topological map and a 2D grid map, and tests were done both indoors and outdoors.…”

Efficient navigation based on the Landmark-Tree map and the Z<inf>&#x221E;</inf> algorithm using an omnidirectional camera

“…introduce a topological map, which triggers a visual servoing mechanism at crucial locations, like narrow passages [11]. Lui and Jarvis present a mobile robot which is equipped with two omnidirectional cameras [12]. The system builds a metric topological map based on Haar-wavelet signatures of unwarped images.…”

Appearance‐based approach to hybrid metric‐topological simultaneous localisation and mapping

“…As regards appearance‐based vision methods, the whole information in the scenes is used. Several alternatives can be used to describe images, such as the principal components of the image [30, 31], Haar wavelet coefficients [32, 33], Fourier components [34, 35], image histograms [36] or image frequency filtering [37]. To complete the information provided by vision systems, it is possible to take also into account the wheel odometry, GPS information or visual odometry and to set up a probabilistic context with all this information.…”