Sunghwan Ahn

Data Association Using Visual Object Recognition for EKF-SLAM in Home Environment

Reliable data association is crucial to localization and map building for mobile robot applications. For that reason, many mobile robots tend to choose vision-based SLAM solutions. In this paper, a SLAM scheme based on visual object recognition, not just a scene matching, in home environment is proposed without using artificial landmarks. For the object-based SLAM, the following algorithms are suggested: 1) a novel local invariant feature extraction by combining advantages of multi-scale Harris corner as a detector and its SIFT descriptor for natural object recognition, 2) the RANSAC clustering for robust object recognition in the presence of outliers and 3) calculating accurate metric information for SLAM update. The proposed algorithms increase robustness by correct data association and accurate observation. Moreover, it also can be easily implemented real-time by reducing the number of representative landmarks, i.e. objects. The performance of the proposed algorithm was verified by experiments using EKF-SLAM with a stereo camera in home-like environments, and it showed that the final pose error was bounded after battery-run-out autonomous navigation for 50 minutes

Robust sonar feature detection for the SLAM of mobile robot

Sonar sensor is an attractive tool for the SLAM of mobile robot because of their economic aspects. This cheap sensor gives relatively accurate range readings if disregarding angular uncertainty and specular reflections. However, these defects make feature detection difficult for the most part of the SLAM. This paper proposes a robust sonar feature detection algorithm. This algorithm gives feature detection methods for both point features and line features. The point feature detection method is based on the TBF (Wijk and Christensen, 2000) scheme. Moreover, three additional processes improve the performance of feature detection as follows; 1) stable intersections; 2) efficient sliding window update; and 3) removal of the false point features on the wall. The line feature detection method is based on the basic property of adjacent sonar sensors. Along the line feature, three adjacent sonar sensors give similar range readings. Using this sensor property, we propose a novel algorithm for line feature detection, which is simple and the feature can be obtained by using only current sensor data. The proposed feature detection algorithm gives a good solution for the SLAM of mobile robots because it gives an accurate feature information for both the point and line features even with sensor errors. Furthermore, a sufficient number of features are available to correct mobile robot pose. Experimental results of the EKF-based SLAM demonstrate the performance of the proposed feature detection algorithm in a home-like environment.

Metric SLAM in Home Environment with Visual Objects and Sonar Features

To increase the intelligence of mobile robot, various sensors need to be fused effectively to cope with uncertainty induced from both environment and sensors. Combining sonar and vision sensors possesses numerous advantages of economical efficiency and complementary cooperation. Especially, it can remedy false data association and divergence problem of sonar sensors, and overcome low frequency vision based SLAM update caused by computational burden and weakness in illumination changes of vision sensors. In this paper, we propose a SLAM method to join sonar sensors and stereo camera together. It consists of two schemes: extracting robust point and line features from sonar data, and recognizing planar visual objects using multi-scale Harris corner detector and its SIFT descriptor from pre-constructed object database. Fusing sonar features and visual objects through EKF-based SLAM can give correct data association via object recognition and high frequency update via sonar features. As a result, it can increase robustness and accuracy of SLAM in home environment. The performance of the proposed algorithm was verified by experiments in home environment with dynamic obstacles

A Practical Solution to SLAM and Navigation in Home Environment

To implement an autonomous mobile robot, both SLAM and task based navigation algorithms should be performed successfully. Especially, the performance of the estimation while the mobile robot performs task based navigation should be guaranteed. For this purpose, we integrate a SLAM method and a navigation algorithm for practical autonomous mobile robot. The SLAM method combines sonar sensors and stereo camera together using the EKF-based SLAM. Fusing sonar features and visual objects can give correct data association via object recognition and high frequency update via sonar features. The navigation algorithm consists of global and local path planner when the goal position is given. The global path planner uses modified A* algorithm and it gives the mobile robot enough opportunity to detect the registered landmarks during moving to the goal position. As a local path planner, for safe obstacle avoidance, we propose circle following (CF) algorithm. The performance of the proposed algorithm was verified by experiments in home environment with dynamic obstacles

SLAM with Visual Plane: Extracting Vertical Plane by Fusing Stereo Vision and Ultrasonic Sensor for Indoor Environment

This paper presents an algorithm for visual SLAM based on a visual plane, a reliable grouping of salient visual features along sonar line features. The grouping of visual features improves data association and reduces the number of landmarks against individual visual features. To accomplish this, we propose three techniques: 1) selection of visual features which are invariant to image changes in indoor environment and suitable candidates for the visual plane, 2) extraction of sonar line features with current sensor data, which filters out uncertain outliers efficiently and 3) a scheme on grouping visual features with respect to sonar line features and maintaining database of the extracted visual planes for reliable data association. We integrate above three techniques into one framework and propose a SLAM algorithm for the visual planes. Experimental results in two types of real home environment show that the algorithm can successfully be executed with no human intervention.

Efficient SLAM algorithm with hybrid visual map in an indoor environment

In this paper, we propose a scheme of generating a hybrid visual map for SLAM. The hybrid visual map has two levels of map representations: 1) the absolute map representation of highly distinctive visual planes via EKF- SLAM and 2) the relative map representation of dense visual features for each visual plane via sparse information filter update. The absolute map can maintain its global consistency by matching the visual plane, the group of visual features. It improves data association and reduces the number of landmarks against individual visual features. Moreover, the relative map can reconstruct a 3-D map of visual features efficiently without loosing dense visual information. The performance of the proposed method was verified by the experimental results of consistent hybrid visual maps in an indoor environment.

Incremental and robust construction of Generalized Voronoi Graph (GVG) for mobile guide robot

GVG has been effectively used as a sensor based navigation tool using 360/spl deg/ sensor data. For mobile guide robot applications, however, we can only use 180/spl deg/ sensor data and the robustness of the navigation algorithm is critical for successful applications. For that purpose, the robot should be equipped with three capabilities. Those are 1) incremental GVG construction, 2) robust GVG navigation and 3) navigation strategy that just uses half of the sensor scan, i.e. 180/spl deg/. In this paper, we propose a GVG navigation algorithm that has above 3 capabilities. We firstly propose a method that can estimate the invisible 180/spl deg/ range from previous range data. Moreover, we suggest a way of robust GVG navigation algorithm by using a sensor data matching technique. The simulation result validates that the proposed algorithm can incrementally and robustly navigate the semi-unstructured map by using 180/spl deg/ sensor scan.

The robust construction of a generalized Voronoi graph (GVG) using partial range data for guide robots

Purpose – The purpose of this paper is to describe research to enable a robust navigation of guide robots in erratic environments with partial sensor information.Design/methodology/approach – Two techniques were developed. One is a robust node discrimination method by using an adaptive sensor matching method. The other is a robot navigation technique with partial sensor information.Findings – A successful navigation was implemented in erratic environments using partial sensor information.Originality/value – First robot navigation is addressed along the generalized Voronoi graph (GVG) with partial sensor information. A solution is also provided for a phantom node detection problem, which is one of the main defects in GVG navigation.

Construction of hybrid visual map for indoor SLAM

Our previous work, SLAM with visual plane, allowed correct data association and computationally feasible solution for vision-based SLAM. In this paper, we propose a scheme of generating a hybrid visual map based on the visual plane framework. The hybrid visual map has two levels of map representations: 1) absolute map representation of distinctive visual planes via EKF-SLAM and 2) relative map representation of dense visual features for each visual plane via sparse information filter update. It can inherit the advantages of the visual plane by maintaining the absolute map. Moreover, the relative map can reconstruct a 3-D map of visual features efficiently without loosing dense visual information. The performance of the proposed method was verified by the experimental results of consistent hybrid visual maps in two real indoor environments.

POSTECH Navigation Frame: Toward a Practical Solution for Indoor SLAM and Navigation

This article addresses SLAM and navigation frame using low-cost sensor systems as a practical solution in an indoor environment. It is divided into three major modules according to their functionality: 1) initial mapping by VR-SLAM (Vision and Range sensor-SLAM) while exploring an unknown environment, 2) task-based navigation to guarantee safe motion under dynamic environmental changes and 3) failure recovery to improve system reliability. These modules work with a stereo camera, ultrasonic and infrared sensors combined. In this article, we will describe the navigation frame, especially, focusing on VR-SLAM.