移动机器人基于三维激光测距的室内场景认知

庄严; 卢希彬; 李云辉

doi:10.3724/SP.J.1004.2011.01232

移动机器人基于三维激光测距的室内场景认知

doi: 10.3724/SP.J.1004.2011.01232

1.
大连理工大学信息与控制研究中心大连 116024

详细信息

通讯作者:
庄严大连理工大学自动化系副教授. 主要研究方向为机器人导航、探索、自主环境建模与环境认知. E-mail: zhuang@dlut.edu.cn

计量
- 文章访问数: 2545
- HTML全文浏览量: 72
- PDF下载量: 1541
- 被引次数: 0
出版历程
- 收稿日期: 2010-11-02
- 修回日期: 2011-03-02
- 刊出日期: 2011-10-20

Mobile Robot Indoor Scene Cognition Using 3D Laser Scanning

1.
Research Center of Information and Control, Dalian University of Technology, Dalian 116024

摘要

摘要: 研究了移动机器人在室内三维环境中的场景认知问题.室内场景框架具有结构化特性,而室内多样化的物体则难以进行模型化表述. 本文利用区域扩张算法进行平面特征的提取,并根据平面属性及其相互间的空间关系,完成室内场景框架的辨识.为了借鉴图像处理领域的物体识别方法, 本文提出一种基于Bearing Angle模型的激光测距数据表述方法,从而将三维点云数据转换为二维Bearing Angle图. 同一类物体中的个体形态具有多样性,同时观测视角也导致激光测距数据的显著差异.针对这些问题,采用一种基于Gentleboost算法的有监督学习方法, 并利用物体碎片及其相对于物体中心的位置作为特征,从而完成室内场景中的物体认知. 利用室内场景框架辨识结果在Bearing Angle图中进行天棚、地面、墙壁、房门等区域的标记,并利用所产生的语义信息去除错误的认知结果,从而有助于提高识别率. 利用实际机器人平台所获得的实验结果验证了所提方法的有效性.
- 三维激光测距 /
- Bearing Angle图 /
- Gentleboost /
- 室内场景认知
Abstract: This paper mainly studies the indoor scene cognition problem for mobile robots. The structure of an indoor scene is assumed to be structured, while indoor objects are in a variety of forms, which are difficult to be represented by specific descriptive models. In our work, planes are extracted from 3D laser data using regional expansion algorithm, and the properties as well as the relationship of these planes are used for the indoor structure identification. In order to adopt digital image processing algorithm to implement object detection, a bearing angle model is used to represent laser point clouds, so that 3D laser scanning data can be converted to 2D bearing angle image. It is difficult to detect a large number of different classes of objects in cluttered indoor scenes, especially when the mobile robot acquires the 3D laser scanning data in different locations and angles of view. An approach based on gentleboost algorithm is proposed for multi-class object detection, which takes the fragments and the location with respect to the object center as the generic features for object detection. As the result of indoor structure identification, the specific region for ceiling, floor, wall and door can be labeled in the bearing angle image. With the help of known semantic information, the false object detection results can be eliminated effectively. Experiment results implemented on a real mobile robot show the validity of the proposed method.
- 3D laser scanning /
- bearing angle image /
- gentleboost /
- indoor scene cognition

HTML全文

参考文献(1)

[1]

Sande K E A, Gevers T, Snoek C G M. Evaluating color descriptors for object and scene recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(9): 1582-1596[2] Felzenszwalb P F, Girshick R B, McAllester D, Ramanan D. Object detection with discriminatively trained part-based models. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(9): 1627-1645[3] Vernaza P, Lee D D. Scalable real-time object recognition and segmentation via cascaded, discriminative Markov random fields. In: Proceedings of the IEEE International Conference on Robotics and Automation. Anchorage, USA: IEEE, 2010. 3102-3107[4] Himmelsbach M, Luettel T, Wuensche H J. Real-time object classification in 3D point clouds using point feature histograms. In: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems. St. Louis, USA: IEEE, 2009. 994-1000[5] Anguelov D, Taskarf B, Chatalbashev V, Koller D, Gupta D, Heitz G, Ng A. Discriminative learning of Markov random fields for segmentation of 3D scan data. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE, 2005. 169-176[6] Agrawal A, Nakazawa A, Takemura H. MMM-classification of 3D range data. In: Proceedings of the IEEE International Conference on Robotics and Automation. Kobe, Japan: IEEE, 2009. 2003-2008[7] Viola P, Jones M. Rapid object detection using a boosted cascade of simple features. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE, 2001. 511-518[8] Torralba A, Murphy K P, Freeman W T. Sharing visual features for multiclass and multiview object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(5): 854-869[9] Nüchter A, Hertzberg J. Towards semantic maps for mobile robots. Robotics and Autonomous Systems, 2008, 56(11): 915-926[10] Surmann H, Lingemann K, Nuchter A, Hertzberg J. Fast acquiring and analysis of three dimensional laser range data. In: Proceedings of the International Fall Workshop Vision, Modeling and Visualization. Stuttgart, Germany: AkaGmbH, 2001. 59-66[11] Stamos I, Allen P K. 3-D model construction using range and image data. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Hilton Head Island, USA: IEEE, 2000. 531-536[12] Scaramuzza D, Harati A, Siegwart R. Extrinsic self calibration of a camera and a 3D laser range finder from natural scenes. In: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems. San Diego, USA: IEEE, 2007. 4164-4169[13] Friedman J, Tibshirani R, Hastie T. Additive logistic regression: a statistical view of boosting. The Annals of Statistics, 2000, 28(2): 337-374[14] Ullman S, Sali E, Vidal-Naquet M. A fragment-based approach to object representation and classification. In: Proceedings of the 4th International Workshop on Visual Form. London, UK: Springer, 2001. 85-102[15] Macqueen J. Some methods for classification and analysis of multivariate observations. In: Proceedings of 5th Berkeley Symposium on Mathematical Statistics and Probability. Berkeley, USA: University of California Press, 1967. 281-297

施引文献

资源附件(0)

访问统计