Feature extraction from laser scan data based on curvature estimation for mobile robotics

Huo

Proceedings of the 33rd Chinese Control Conference

2014

In this paper, we present a real-time 3D robotic indoor mapping system on a wheeled mobile robot equipped with two 2D laser range finder (LRF). We propose the fusion approach of two fixed 2D laser range finder (one horizontal and one vertical), odometers and gyroscope, to build accurate 2D and 3D maps at the same time. The two LRFs perform line-feature based SLAM in horizontal and vertical planes when the robot walking forward. Combined with the pose computed by the information of odometers and gyroscope, we achieve an accurate localization. To represent the 3D environment an octree based map is used. Mapping experiment of a 180m+ corridor was tested on an average speed of 0.5m/s without stop. The results shows the efficiency and accuracy of the 3D mapping system.

Section: Line Feature Extractionmentioning

confidence: 99%

Section: Line Feature Extractionmentioning

confidence: 99%

A real-time robotic indoor 3D mapping system using duel 2D laser range finders

Huo

Proceedings of the 33rd Chinese Control Conference

2014

2013 IEEE International Symposium on Robotic and Sensors Environments (ROSE)

“…Mobile robot systems usually detect such geometric features from range data sampled from the environment [4], [5], [6], [7], [1], [8], [9]. Most of these approaches use features, e.g.…”

Section: Feature Definitionmentioning

confidence: 99%

High-speed and accurate laser scan matching using classified features

Shu

Huang

2013

Laser scan matching algorithm plays a key role in robot localization and mapping. In this paper, we propose a classified feature-based algorithm that matches laser scans in a closed-form manner called Classified Feature-based Scan Matcher (CFSM). Based on a geometric observation, our classified features are defined as rotational features and translational features separately to improve matching accuracy. Experimental results demonstrate that CFSM can produce better accuracy for scans with large angular displacement, without increasing running time. Indoor robot can take advantage of this algorithm in performing fast and accurate pose estimation.

2011 IEEE International Conference on Robotics and Automation

“…Classical feature detectors exploit the characteristics of an environment: in indoor settings, lines, corners and curves have been used [10], [11], [12], [13], [14]. Outdoors, the hand-tuned tree detector originally developed for the Victoria Park dataset [15], has been used almost universally (see [16], [8], [17] for representative examples).…”

Section: Introductionmentioning

confidence: 99%

Structure tensors for general purpose LIDAR feature extraction

Olson

2011

Abstract-The detection of features from Light Detection and Ranging (LIDAR) data is a fundamental component of featurebased mapping and SLAM systems. Classical approaches are often tied to specific environments, computationally expensive, or do not extract precise features.We describe a general purpose feature detector that is not only efficient, but also applicable to virtually any environment. Our method shares its mathematical foundation with feature detectors from the computer vision community, where structure tensor based methods have been successful. Our resulting method is capable of identifying stable and repeatable features at a variety of spatial scales, and produces uncertainty estimates for use in a state estimation algorithm. We verify the proposed method on standard datasets, including the Victoria Park dataset and the Intel Research Center dataset.