Multi-hypothesis localization with a rough map using multiple visual features for outdoor navigation

Abstract

We describe a method of mobile robot localization based on a rough map using stereo vision, which uses multiple visual features to detect and segment the buildings in the robot's field of view. The rough map is an inaccurate map with large uncertainties in the shapes, dimensions and locations of objects so that it can be built easily. The robot fuses odometry and vision information using extended Kalman filters to update the robot pose and the associated uncertainty based on the recognition of buildings in the map. We use a multi-hypothesis Kalman filter to generate and track Gaussian pose hypotheses. An experimental result shows the feasibility of our localization method in an outdoor environment.

A Quantitative Navigability Measure of Rough Maps

“…This section briefly reviews our multi-hypothesis localization method using a rough map. Refer to our previous paper [11] for more details.…”

“…The simulated robot move away from the starting position, turns at the subgoals and stops at the destination position in the figure. Since we are using our previous method of multi-hypothesis localization [11], the detailed strategy of the simulated motion is as follows: When the robot recognized a subgoal or a destination, its pose hypothesis would be reserved. When the robot failed to recognize a goal, it travels over more distances searching for the goal.…”

A quantitative measure for the navigability of a mobile robot using rough maps

“…We use our Multi-Hypothesis Localization (MHL) method using a rough map for analyzing the robot behavior in simulated environments [10]. The simulation system includes a representation of the environment, a vision simulator, and a motion simulator.…”

Outdoor visual localization with a hand-drawn line drawing map using FastSLAM with PSO-based mapping

“…Yun and Miura [22] proposed a localization method using a line drawing building map with uncertainty. They used line segments in the image and vertical planes in stereo data as features for localization.…”