[1]陈焕朝,田国会.一种服务机器人选择性认知方法[J].东南大学学报(自然科学版),2019,49(1):82-87.[doi:10.3969/j.issn.1001-0505.2019.01.012]
 Chen Huanzhao,Tian Guohui.Selective recognition method for service robots[J].Journal of Southeast University (Natural Science Edition),2019,49(1):82-87.[doi:10.3969/j.issn.1001-0505.2019.01.012]
点击复制

一种服务机器人选择性认知方法()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
49
期数:
2019年第1期
页码:
82-87
栏目:
计算机科学与工程
出版日期:
2019-01-20

文章信息/Info

Title:
Selective recognition method for service robots
作者:
陈焕朝田国会
山东大学控制科学与工程学院, 济南 250061
Author(s):
Chen Huanzhao Tian Guohui
School of Control Science and Engineering, Shandong University, Jinan 250061, China
关键词:
视觉注意 目标检测 区域卷积神经网络 语义分析
Keywords:
visual attention object detection regions with convolutional neural network semantic analysis
分类号:
TP391
DOI:
10.3969/j.issn.1001-0505.2019.01.012
摘要:
针对现阶段服务机器人面向日常家居环境的认知能力较低,缺乏以相关服务任务作为先验知识、进行深度认知的能力问题,基于视觉认知和语义分析相结合的思路,利用智能空间平台分布式感知系统,结合深度神经网络下图像处理与语义分析等相关手段,提出了一种服务机器人选择性语义认知方法.设计了一种智能空间下选择性注意策略,选择显著性高、服务任务相关联的区域作为兴趣区域;提出了一种改进的区域卷积神经网络,将所选择兴趣区域作为先验知识来引导卷积神经网络;构建了语义分析模型,利用物品检测输出进行深度语义认知.实验结果表明,该方法可以提取服务任务相关的空间显著性信息,引导区域卷积神经网络进行物品检测,物品检测过程过滤了无关物品,与传统区域卷积神经网络相比,检测速度较快、准确率高,最终结合语义分析的方法提高了服务机器人日常家居环境下的认知能力.
Abstract:
Aimed at that the service robots have a low cognitive ability for the daily home environment, and lack the ability to transfer the service tasks to a priori knowledge in deep cognition. The distributed sensing system of an intelligent space platform was used. The image processing and the semantic analysis based on a deep neural network were combined. A selective recognition method for service robots is proposed. A selective attention method in intelligent space was proposed. The areas with high significance and service tasks as the focus of attention(FOA)were picked out. An improved regional convolutional neural network was proposed. The FOA was transferred into a priori knowledge to guide the convolutional neural network for identification. A semantic analysis model was established. The deep knowledge was generated by the semantic analysis model. The results show that the method can extract saliency information related to the service task. Regions with neural networks are guided to generate the deep knowledge about objects. Redundant objects are eliminated during the detecting process. Compared with the traditional approach based on regional convolutional neural network, the method has advantages of the faster detection speed and the higher accuracy.The cognitive ability of service robots in the home environment is improved by the semantic analysis operation.

参考文献/References:

[1] Johnson J,Hariharan B, Maaten L V D, et al. Inferring and executing programs for visual reasoning[C]// IEEE International Conference on Computer Vision.Venice, Italy, 2017: 3008-3017. DOI:10.1109/ICCV.2017.325.
[2] 王田苗, 陶永, 陈阳. 服务机器人技术研究现状与发展趋势[J]. 中国科学:信息科学, 2012, 42(9):1049-1066. DOI:10.1360/112012-402.
Wang T M, Tao Y, Chen Y M. Research status and development trends of the service robotic technology [J]. Scientia Sinica Informationis, 2012, 42(9): 1049-1066. DOI:10.1360/112012-402. (in Chinese)
[3] Lee J H, Hashimoto H. Intelligent space: Concept and contents[J]. Advanced Robotics, 2002, 16(3): 265-280. DOI:10.1163/156855302760121936.
[4] Itti L, Koch C, Niebur E. A model of saliency-based visual attention for rapid scene analysis[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998, 20(11): 1254-1259. DOI:10.1109/34.730558.
[5] Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//IEEE Conference on Computer Vision and Pattern Recognition.Columbus, OH, USA, 2014: 580-587. DOI:10.1109/CVPR.2014.81.
[6] Liu W,Anguelov D, Erhan D, et al. SSD: Single shot multibox detector[C]// European Conference on Computer Vision. Amsterdam, The Netherlands, 2016: 21-37. DOI:10.1007/978-3-319-46448-0_2.
[7] Redmon J, Farhadi A. YOLO9000: Better, faster, stronger[C]// IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA, 2017: 6517-6525. DOI:10.1109/CVPR.2017.690.
[8] Girshick R. Fast R-CNN [C]// IEEE International Conference on Computer Vision. Santiago, Chile, 2015: 1440-1448.DOI:10.1109/ICCV.2015.169.
[9] Ren S Q, He K M, Girshick R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149. DOI:10.1109/tpami.2016.2577031.
[10] Krause J, Johnson J, Krishna R,et al. A Hierarchical approach for generating descriptive image paragraphs[C]// IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA, 2017:3337-3345. DOI:10.1109/CVPR.2017.356.
[11] Zhu Y K, Gordon D, Kolve E, et al. Visual semantic planning using deep successor representations[C]// IEEE International Conference on Computer Vision.Venice, Italy, 2017: 483-492. DOI:10.1109/ICCV.2017.60.
[12] 田国会, 尹建芹, 闫云章, 等. 基于混合高斯模型和主成分分析的轨迹分析行为识别方法[J]. 电子学报, 2016, 44(1): 143-149. DOI:10.3969/j.issn.0372-2112.2016.01.021.
Tian G H, Yin J Q, Yan Y Z, et al. Gaussian mixture models and principal component analysis based human trajectory behavior recognition[J]. Acta Electronica Sinica, 2016, 44(1): 143-149. DOI:10.3969/j.issn.0372-2112.2016.01.021. (in Chinese)

相似文献/References:

[1]陈鹏,吴乐南.压缩感知在多载波EBPSK雷达中的应用[J].东南大学学报(自然科学版),2014,44(1):23.[doi:10.3969/j.issn.1001-0505.2014.01.005]
 Chen Peng,Wu Lenan.Compressive sensing for multi-carrier EBPSK radar[J].Journal of Southeast University (Natural Science Edition),2014,44(1):23.[doi:10.3969/j.issn.1001-0505.2014.01.005]

备注/Memo

备注/Memo:
收稿日期: 2018-07-20.
作者简介: 陈焕朝(1988—),博士生;田国会(联系人),男,博士,教授, 博士生导师,g.h.tian@sdu.edu.cn.
基金项目: 国家自然科学基金资助项目(U1813215,61773239)、山东省泰山学者工程资助项目.
引用本文: 陈焕朝,田国会.一种服务机器人选择性认知方法[J].东南大学学报(自然科学版),2019,49(1):82-87. DOI:10.3969/j.issn.1001-0505.2019.01.012.
更新日期/Last Update: 2019-01-20