·682· 智能系统学报 第16卷 an Conference on Robotics and Automati-on.Melbourne, [20]CUMMINS M.NEWMAN P.FAB-MAP:Probabilistic Australasian.2014:8-14 localization and mapping in the space of appearance[J]. [9]SUNDERHAUF N.SHIRAZI S.DAYOUB F,et al.On The international journal of robotics research,2008. the performance of ConvNet features for place 27(6:647-665 recognition[C]//Proceedings of 2015 IEEE/RSJ Interna- [21]CHOI Y,KIM N,PARK K,et al.All-day visual place re- tional Conference on Intelligent Robots and Systems. cognition:benchmark dataset and baseline[C]//Proceed- Hamburg,Germany,2015:4297-4304. ings of 2015 IEEE International Conference on Computer [10]GAO Xiang,ZHANG Tao.Unsupervised learning to de- Vision and Pattern Recognition Workshops.Boston, tect loops using deep neural networks for visual SLAM USA,2015:8-13. system[J].Autonomous robots,2017,41(1):1-18. [22]KRIZHEVSKY A.SUTSKEVER I.HINTON G E.Im- [11]GOODFELLOW I J.POUGET-ABADIE J.MIRZA M. ageNet classification with deep convolutional neural net- et al.Generative adversarial nets[Cl//Proceedings of the works[C]//Proceedings of the 25th International Confer- 27th International Conference on Neural Information Pro- ence on Neural Information Processing Systems.Lake cessing Systems.Montreal,Canada,2014:2672-2680. Tahoe,USA,2012:1097-1105. [12]HINTON G E.ZEME R S.Autoencoders,minimum de- [23]CHEN Zetao,JACOBSON A,SUNDERHAUF N,et al. scription length and Helmholtz free energy[Cl//Proceed- Deep learning features at scale for visual place recogni- ings of the 6th International Conference on Neural In- tion[C]//Proceedings of 2017 IEEE International Confer- formation Processing Systems.Denver,Colorado,USA. ence on Robotics and Automation.Singapore,2017: 1993:3-10. 3223-3230. [13]SMOLENSKY P.Information processing in dynamical [24]ZAFFAR M,KHALIQ A,EHSAN S,et al.Levelling the systems:foundations of harmony theory[M]//RUMEL- playing field:A comprehensive comparison of visual HART D E,MCCLELLAND J L.Parallel Distributed place recognition approaches under changing Processing:Explorations in the Microstructure of Cogni- conditions[EB/0L].(2019-04-29)[2020-02-01] https://arxiv.org/abs/1903.09107?context=cs.CV. tion.Cambridge:MIT Press,1986. [14]SHIN D W,HO Y S,KIM E S.Loop closure detection in [25]MEMON A R,WANG Hesheng,HUSSAIN A.Loop simultaneous localization and mapping using descriptor closure detection using supervised and unsupervised deep neural networks for monocular SLAM systems[J].Robot- from generative adversarial network[J].Journal of elec- ics and autonomous systems,2020,126:103470. tronic imaging,2019,28(1):013014. [15]RADFORD A,METZ L,CHINTALA S.Unsupervised 作者简介： representation learning with deep convolutional generat- 杨慧，硕土研究生，主要研究方向 ive adversarial networks[C]//Proceedings of the 4th Inter- 为视觉回环检测。 national Conference on Learning Representations.San Juan,Puerto Rico,2016:97-108. [16]SALIMANS T,GOODFELLOW I,ZAREMBA W,et al. Improved techniques for training GANs[Cl//Proceedings of the 30th International Conference on Neural Informa- tion Processing Systems.Barcelona,Spain,2016: 陈良，副教授，主要研究方向为基 2234-2242. 于深度学习的人工智能系统、新一代 [17]DONG Haowen,YANG Y H.Training generative ad- 智能控制理论及应用。 versarial networks with binary neurons by end-to-end backpropagation[EB/OL].(2018-12-12)[2020-01-01]ht- tps://arxiv.org/abs/1810.04714. [18]CAO Yanshuai,DING G W,LUI K Y C,et al.Improv- ing GAN training via binarized representation entropy 孙立宁，教授博士生导师，主要 (BRE)regularization[C]//Proceedings of the 6th Interna- 研究方向为先进机器人技术。主持 tional Conference on Learning Representations.Van- “863”计划、973计划、国家重大专项、 couver,Canada,2018:1-22. 国家自然科学基金等20多项。获国 [19]ZHOU Bolei,LAPEDRIZA A,KHOSLA A,et al.Places: 家技术发明/科技进步二等奖2项、教 A 10 million image database for scene recognition[J]. 育部技术发明奖二等奖1项、省级技 IEEE transactions on pattern analysis and machine intelli- 术发明/科技进步一等奖3项，二等奖 gence,2018,40(6):1452-1464 2项。发表学术论文400多篇，获授权国家发明专利40余项。an Conference on Robotics and Automati-on. Melbourne, Australasian, 2014: 8−14. SÜNDERHAUF N, SHIRAZI S, DAYOUB F, et al. On the performance of ConvNet features for place recognition[C]//Proceedings of 2015 IEEE/RSJ Interna￾tional Conference on Intelligent Robots and Systems. Hamburg, Germany, 2015: 4297−4304. [9] GAO Xiang, ZHANG Tao. Unsupervised learning to de￾tect loops using deep neural networks for visual SLAM system[J]. Autonomous robots, 2017, 41(1): 1–18. [10] GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]//Proceedings of the 27th International Conference on Neural Information Pro￾cessing Systems. Montreal, Canada, 2014: 2672−2680. [11] HINTON G E, ZEME R S. Autoencoders, minimum de￾scription length and Helmholtz free energy[C]//Proceed￾ings of the 6th International Conference on Neural In￾formation Processing Systems. Denver, Colorado, USA, 1993: 3−10. [12] SMOLENSKY P. Information processing in dynamical systems: foundations of harmony theory[M]//RUMEL￾HART D E, MCCLELLAND J L. Parallel Distributed Processing: Explorations in the Microstructure of Cogni￾tion. Cambridge: MIT Press, 1986. [13] SHIN D W, HO Y S, KIM E S. Loop closure detection in simultaneous localization and mapping using descriptor from generative adversarial network[J]. Journal of elec￾tronic imaging, 2019, 28(1): 013014. [14] RADFORD A, METZ L, CHINTALA S. Unsupervised representation learning with deep convolutional generat￾ive adversarial networks[C]//Proceedings of the 4th Inter￾national Conference on Learning Representations. San Juan, Puerto Rico, 2016: 97−108. [15] SALIMANS T, GOODFELLOW I, ZAREMBA W, et al. Improved techniques for training GANs[C]//Proceedings of the 30th International Conference on Neural Informa￾tion Processing Systems. Barcelona, Spain, 2016: 2234−2242. [16] DONG Haowen, YANG Y H. Training generative ad￾versarial networks with binary neurons by end-to-end backpropagation[EB/OL]. (2018-12-12) [2020-01-01] ht￾tps://arxiv.org/abs/1810.04714. [17] CAO Yanshuai, DING G W, LUI K Y C, et al. Improv￾ing GAN training via binarized representation entropy (BRE) regularization[C]//Proceedings of the 6th Interna￾tional Conference on Learning Representations. Van￾couver, Canada, 2018: 1−22. [18] ZHOU Bolei, LAPEDRIZA A, KHOSLA A, et al. Places: A 10 million image database for scene recognition[J]. IEEE transactions on pattern analysis and machine intelli￾gence, 2018, 40(6): 1452–1464. [19] CUMMINS M, NEWMAN P. FAB-MAP: Probabilistic localization and mapping in the space of appearance[J]. The international journal of robotics research, 2008, 27(6): 647–665. [20] CHOI Y, KIM N, PARK K, et al. All-day visual place re￾cognition: benchmark dataset and baseline[C]//Proceed￾ings of 2015 IEEE International Conference on Computer Vision and Pattern Recognition Workshops. Boston, USA, 2015: 8−13. [21] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Im￾ageNet classification with deep convolutional neural net￾works[C]//Proceedings of the 25th International Confer￾ence on Neural Information Processing Systems. Lake Tahoe, USA, 2012: 1097−1105. [22] CHEN Zetao, JACOBSON A, SÜNDERHAUF N, et al. Deep learning features at scale for visual place recogni￾tion[C]//Proceedings of 2017 IEEE International Confer￾ence on Robotics and Automation. Singapore, 2017: 3223−3230. [23] ZAFFAR M, KHALIQ A, EHSAN S, et al. Levelling the playing field: A comprehensive comparison of visual place recognition approaches under changing conditions[EB/OL]. (2019-04-29) [2020-02-01] https://arxiv.org/abs/1903.09107?context=cs.CV. [24] MEMON A R, WANG Hesheng, HUSSAIN A. Loop closure detection using supervised and unsupervised deep neural networks for monocular SLAM systems[J]. Robot￾ics and autonomous systems, 2020, 126: 103470. [25] 作者简介： 杨慧，硕士研究生，主要研究方向 为视觉回环检测。 陈良，副教授，主要研究方向为基 于深度学习的人工智能系统、新一代 智能控制理论及应用。 孙立宁，教授，博士生导师，主要 研究方向为先进机器人技术。主持 “863”计划、973 计划、国家重大专项、 国家自然科学基金等 20 多项。获国 家技术发明/科技进步二等奖 2 项、教 育部技术发明奖二等奖 1 项、省级技 术发明/科技进步一等奖 3 项，二等奖 2 项。发表学术论文 400 多篇，获授权国家发明专利 40 余项。 ·682· 智 能 系 统 学 报 第 16 卷