Path planning for human beings is based on scene understanding and semantic map as well as geometry features. The point cloud map can help robots to perform path planning. However, it is quite different from the natural way of human beings. In this paper, we propose to construct a novel framework for topological semantic map. Specifically, we construct a 2D semantic map by projecting 3D scene semantic information recognized by convolutional neural network onto a 2D plane. 3D reconstruction of the environment is achieved by RGB-D SLAM 3D space mapping algorithm. The intersections in the 2D map are recognized offline, and the semantic annotation of the intersection in the topological map is utilized to build up a complete object-based semantic map. Experimental results demonstrate the feasibility of the proposed approach.
Mur-ArtalR., MontielJ.M.M. and TardósJ.D., ORB-SLAM: Aversatile and Accurate Monocular SLAM System, IEEE Transactionson Robotics31 (2017), 1147–1163.
2.
EngelJ., SchöpsT. and CremersD., LSD-SLAM: Large-Scale DirectMonocular SLAM, in European Conference on Computer Vision2014, pp. 834–849.
3.
KostavelisI. and GasteratosA., Semantic mapping for mobile robotics tasks: A survey, Autonomous Systems66 (2015) 86–103.
4.
LowryS., SünderhaufN., NewmanP.LeonardJ.J., CoxD., CorkeP., et al., Visual place recognition: A survey, IEEE Transactions on Robotics32 (2016), 1–19.
5.
NiuJ., BuX., QianK. and LiZ., An indoor scene recognition method combining global and saliency region features, Jiqiren/robot37 (2015), 122–128.
6.
LiX. and BelaroussiR., i-dense 3D semantic mapping from monocular SLAM, 2016.
7.
MozosO.M., StachnissC. and BurgardW., Supervised learning of places from range data using adaboost , in Robotics and Automation, 2005. ICRA 2005. Proceedings of the 2005 IEEE International Conference on 2005, pp. 1730–1735.
8.
KondaK. and MemisevicR., Learning visual odometry with a convolutional network , in International Conference on Computer Vision Theory and Applications2015, pp. 486–490.
9.
DosovitskiyA., FischeryP., IlgE. and HäusserP., FlowNet:Learning optical flow with convolutional networks, in IEEE International Conference on Computer Vision2015, pp. 2758–2766.
10.
ShahidM., NaseerT. and BurgardW., DTLC: Deeply trained loopclosure detections for lifelong visual SLAM, Visual PlaceRecognition: What is it Good for2016.
11.
GaoX. and ZhangT., Loop closure detection for visual SLAM systemsusing deep neural networks, in Control Conference2015, pp. 5851–5856.
12.
FischerP., DosovitskiyA. and BroxT., Descriptor matching with convolutional neural networks: A comparison to sift, arXiv preprint arXiv:1405.5769,2014.
13.
GirshickR., DonahueJ., DarrellT. and MalikJ., Region-based convolutional networks for accurate object detection and segmentation, IEEE Transactions on Pattern Analysis & Machine Intelligence38 (2015), 142–158.
14.
Salas-MorenoR.F., Dense semantic SLAM, 2014.
15.
McCormacJ., HandaA., DavisonA. and LeuteneggerS., Semanticfusion: Dense 3d semantic mapping with convolutional neural networks, arXiv preprint arXiv:1609.05130,2016.
16.
TenorthM., KunzeL., JainD. and BeetzM., Knowrob-map-knowledge-linked semantic object maps, in Humanoid Robots (Humanoids), 2010 10th IEEE-RAS International Conference on2010, pp. 430–435.
17.
TrevorA.J., GedikliS., RusuR.B. and ChristensenH.I., efficient organized point cloud segmentation with connected components, Semantic Perception Mapping and Exploration (SPME)2013.
18.
SwadzbaA. and WachsmuthS., A detailed analysis of a new 3D spatial feature vector for indoor scene classification, Robotics and Autonomous Systems62 (2014), 646–662.
19.
BuschkaP. and SaffiottiA., A virtual sensor for room detection, in Intelligent Robots and Systems, 2002. IEEE/RSJ International Conference on2002, pp. 637–642.
20.
EspinaceP., KollarT., RoyN. and SotoA., Indoor scene recognition by a mobile robot through adaptive object detection, Robotics and Autonomous Systems61 (2013), 932–947.
21.
SünderhaufN., DayoubF., McMahonS., TalbotB., SchulzR.CorkeP., et al., Place categorization and semantic mapping on a mobilerobot, inRobotics and Automation (ICRA), 2016 IEEE International Conference on2016, pp. 5729–5736.
22.
GoeddelR. and OlsonE., Learning semantic place labels fromoccupancy grids using CNNs, in Intelligent Robots and Systems(IROS), 2016 IEEE/RSJ International Conference on2016, pp. 3999–4004.
23.
WangL., GuoS., HuangW. and QiaoY., Places205-vggnet models for scene recognition, arXiv preprint arXiv: 1508.016672015.
24.
SimonyanK. and ZissermanA., Very deep convolutional networks forlarge-scale image recognition, arXiv preprint arXiv:1409.15562014.
25.
EndresF., HessJ., EngelhardN., SturmJ., CremersD. and BurgardW., An evaluation of the RGB-D SLAM system, in Roboticsand Automation (ICRA), 2012 IEEE International Conference on2012, pp 1691–1696.
