·90· 智能系统学报 第16卷 限的问题。实验结果表明，本文所提出的室外图 估计).华中科技大学学报（自然科学版），2020,48(5) 像深度估计方法在Make3D标准数据集和实际拍 7-12 摄的室外单帧图像上都取得了较高的深度估计精 WANG Quande,ZHANG Songtao.Monocular depth es- 度。在此基础上将其与基于视觉的移动机器人坡 timation with multi-scale feature fusion[J].Journal of 度检测相结合进一步提出了一种改进的移动机器 Huazhong University of Science and Technology (natural 人单帧图像坡度检测算法，实验结果表明移动机 science edition),2020,48(5):7-12. 器人坡度检测精度得到了显著提高，检测精度满 [8]毕天腾，刘越，翁冬冬，等.基于监督学习的单幅图像深 足移动机器人在未知环境中对坡度感知精度的要 度估计综述[).计算机辅助设计与图形学学报，2018， 求。在今后的工作中，我们将进一步将本文所提 30(8):1383-1393 出的单帧图像深度估计结果应用到图像语义分割 BI Tianteng,LIU Yue,WENG Dongdong,et al.Survey on 中以提高图像语义分割网络的性能，最终完成未 supervised learning based depth estimation from a single 知环境下的高精度3D语义地图构建。 image[J].Journal of computer-aided design and computer graphics,2018.30(8):1383-1393 [9]何磊，苏松志，李绍滋.单摄像头下基于样本学习的人体 参考文献： 深度估计U.智能系统学报，2014,9(2)：161-167 [1]朱江，王耀南，余洪山，等.未知环境下移动机器人自主 HE Lei,SU Songzhi,LI Shaozi.Human depth estimation 感知斜坡地形方法[.仪器仪表学报，2010,31(8)月 on the basis of the sample learning method under a single 1916-1920 camera[J].CAAI transactions on intelligent systems,2014, ZHU Jiang,WANG Yaonan,YU Hongshan,et al.Mobile 9(2:161-167 robot autonomous perceiving slope terrain under unknown [10]孙蕴瀚，史金龙，孙正兴.利用自监督卷积网络估计单 environment[J].Chinese journal of scientific instrument, 图像深度信息[】.计算机辅助设计与图形学学报， 2010,31(8):1916-1920. 2020,32(4):643-651 [2]YU Jinxia,CAI Zixing,DUAN Zhuohua.Dead reckoning SUN Yunhan,SHI Jinlong,SUN Zhengxing.Estimating of mobile robot in complex terrain based on propriocept- depth from single image using unsupervised convolution- ive sensors[C]//Proceedings of 2008 International Confer- al network[J].Journal of computer-aided design and com- ence on Machine Learning and Cybernetics.Kunming, puter graphics,2020,32(4):643-651. China,2008:1930-1935 [11]EIGEN D,PUHRSCH C,FERGUS R.Depth map predic- [3]LI Zhibin,TSAGARAKIS N G,CALDWELL D G.Stabil- tion from a single image using a multi-scale deep net- izing humanoids on slopes using terrain inclination estima- work[Cl//Proceedings of the 27th International Confer- tion[C]//Proceedings of 2013 IEEE/RSJ International Con- ence on Neural Information Processing Systems. ference on Intelligent Robots and Systems.Tokyo,Japan, Montreal,Quebec,Canada,2014:2366-2374. 2013:4124-4129 [12]LIU Fayao,SHEN Chunhua,LIN Guosheng,et al.Learn- [4]LU Jixin,KOBAYASHI Y,EMARU T,et al.Indoor slope ing depth from single monocular images using deep con- and edge detection by using two-dimensional EKF-SLAM volutional neural fields[J].IEEE transactions on pattern with orthogonal assumption[J].International journal of ad- analysis and machine intelligence,2016,38(10): vanced robotic systems,2015,12(4):44. 2024-2039 [5]HARA S.SHIMIZU T.KONISHI M,et al.Autonomous [13]RANFTL R.LASINGER K,HAFNER D,et al.Towards mobile robot for outdoor slope using 2D LiDAR with uni- robust monocular depth estimation:mixing datasets for axial gimbal mechanism[J].Journal of robotics and zero-shot cross-dataset transfer[J.IEEE transactions on mechatronics,.2020,32(6):1173-1182. pattern analysis and machine intelligence,2020,42(8): [6]TAREEN S A K,KHAN H M.Novel slope detection and 1939-3539. calculation techniques for mobile robots[C]//Proceedings [14]HU Junjie,OZAY M,ZHANG Yan,et al.Revisiting of the 2nd IEEE International Conference on Robotics single image depth estimation:toward higher resolution and Artificial Intelligence.Rawalpindi,Pakistan,2016: maps with accurate object boundaries[J].arXiv:1803. 158-163. 0867,2018 [7]王泉德，张松涛.基于多尺度特征融合的单目图像深度 [15]KUZNIETSOV Y.STUCKLER J,LEIBE B.Semi-super限的问题。实验结果表明，本文所提出的室外图 像深度估计方法在 Make 3D 标准数据集和实际拍 摄的室外单帧图像上都取得了较高的深度估计精 度。在此基础上将其与基于视觉的移动机器人坡 度检测相结合进一步提出了一种改进的移动机器 人单帧图像坡度检测算法，实验结果表明移动机 器人坡度检测精度得到了显著提高，检测精度满 足移动机器人在未知环境中对坡度感知精度的要 求。在今后的工作中，我们将进一步将本文所提 出的单帧图像深度估计结果应用到图像语义分割 中以提高图像语义分割网络的性能，最终完成未 知环境下的高精度 3D 语义地图构建。 参考文献： 朱江, 王耀南, 余洪山, 等. 未知环境下移动机器人自主 感知斜坡地形方法 [J]. 仪器仪表学报, 2010, 31(8): 1916–1920. ZHU Jiang, WANG Yaonan, YU Hongshan, et al. Mobile robot autonomous perceiving slope terrain under unknown environment[J]. Chinese journal of scientific instrument, 2010, 31(8): 1916–1920. [1] YU Jinxia, CAI Zixing, DUAN Zhuohua. Dead reckoning of mobile robot in complex terrain based on propriocept￾ive sensors[C]//Proceedings of 2008 International Confer￾ence on Machine Learning and Cybernetics. Kunming, China, 2008: 1930–1935. [2] LI Zhibin, TSAGARAKIS N G, CALDWELL D G. Stabil￾izing humanoids on slopes using terrain inclination estima￾tion[C]//Proceedings of 2013 IEEE/RSJ International Con￾ference on Intelligent Robots and Systems. Tokyo, Japan, 2013: 4124–4129. [3] LU Jixin, KOBAYASHI Y, EMARU T, et al. Indoor slope and edge detection by using two-dimensional EKF-SLAM with orthogonal assumption[J]. International journal of ad￾vanced robotic systems, 2015, 12(4): 44. [4] HARA S, SHIMIZU T, KONISHI M, et al. Autonomous mobile robot for outdoor slope using 2D LiDAR with uni￾axial gimbal mechanism[J]. Journal of robotics and mechatronics, 2020, 32(6): 1173–1182. [5] TAREEN S A K, KHAN H M. Novel slope detection and calculation techniques for mobile robots[C]//Proceedings of the 2nd IEEE International Conference on Robotics and Artificial Intelligence. Rawalpindi, Pakistan, 2016: 158–163. [6] [7] 王泉德, 张松涛. 基于多尺度特征融合的单目图像深度 估计 [J]. 华中科技大学学报(自然科学版), 2020, 48(5): 7–12. WANG Quande, ZHANG Songtao. Monocular depth es￾timation with multi-scale feature fusion[J]. Journal of Huazhong University of Science and Technology (natural science edition), 2020, 48(5): 7–12. 毕天腾, 刘越, 翁冬冬, 等. 基于监督学习的单幅图像深 度估计综述 [J]. 计算机辅助设计与图形学学报, 2018, 30(8): 1383–1393. BI Tianteng, LIU Yue, WENG Dongdong, et al. Survey on supervised learning based depth estimation from a single image[J]. Journal of computer-aided design and computer graphics, 2018, 30(8): 1383–1393. [8] 何磊, 苏松志, 李绍滋. 单摄像头下基于样本学习的人体 深度估计 [J]. 智能系统学报, 2014, 9(2): 161–167. HE Lei, SU Songzhi, LI Shaozi. Human depth estimation on the basis of the sample learning method under a single camera[J]. CAAI transactions on intelligent systems, 2014, 9(2): 161–167. [9] 孙蕴瀚, 史金龙, 孙正兴. 利用自监督卷积网络估计单 图像深度信息 [J]. 计算机辅助设计与图形学学报, 2020,32(4): 643–651. SUN Yunhan, SHI Jinlong, SUN Zhengxing. Estimating depth from single image using unsupervised convolution￾al network[J]. Journal of computer-aided design and com￾puter graphics, 2020,32(4): 643–651. [10] EIGEN D, PUHRSCH C, FERGUS R. Depth map predic￾tion from a single image using a multi-scale deep net￾work[C]//Proceedings of the 27th International Confer￾ence on Neural Information Processing Systems. Montreal, Quebec, Canada, 2014: 2366–2374. [11] LIU Fayao, SHEN Chunhua, LIN Guosheng, et al. Learn￾ing depth from single monocular images using deep con￾volutional neural fields[J]. IEEE transactions on pattern analysis and machine intelligence, 2016, 38(10): 2024–2039. [12] RANFTL R, LASINGER K, HAFNER D, et al. Towards robust monocular depth estimation: mixing datasets for zero-shot cross-dataset transfer[J]. IEEE transactions on pattern analysis and machine intelligence, 2020, 42(8): 1939–3539. [13] HU Junjie, OZAY M, ZHANG Yan, et al. Revisiting single image depth estimation: toward higher resolution maps with accurate object boundaries[J]. arXiv: 1803. 0867, 2018. [14] [15] KUZNIETSOV Y, STÜCKLER J, LEIBE B. Semi-super- ·90· 智 能 系 统 学 报 第 16 卷