LOCALIZATION OF MOBILE ROBOT WITH PARTICLE FILTER AT DETECTION AND SEGMENTATION OF OBJECTS

Subject of Research. The paper presents a method for mobile robot localization using a particle filter (Monte-Carlo method) based on computer vision. The localization algorithm uses distinctive landmarks that are understandable to a man. Semantic information is used in the motion model with and without range data. Method. The operation principle of the modified localization algorithm lies in applying high-level semantic prompts. Instead of squeezing the floor plan into the third dimension, the three-dimensional world is convolved into a two-dimensional representation and a sample of high-level discriminatory landmarks. This approach is used to represent global localization, which relies exclusively on semantic labels present in the floor plan and extracted from RGB images. Main Results. We demonstrate that localization with segmentation of objects, based on distinctive landmarks, is an effective alternative to traditional scanning. The study is performed in a floor plan data set, and several approaches are compared in terms of qualitative and quantitative localization at room level and global localization. It is shown that semantic information complements modern methods, ensuring that errors are reduced to 35 %. Practical Relevance. We have presented a new structure of perception and localization which uses semantic data and information about distances. The new platform can be used for localization as superior to traditional algorithms based on the Monte Carlo method.

1. Liu C., Schwing A.G., Kundu K., Urtasun R., Fidler S. Rent3D: Floor-plan priors for monocular layout estimation. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 2015. doi: 10.1109/CVPR.2015.7298963
2. Thrun S., Fox D., Burgard W., Dellaert F. Robust Monte Carlo localization for mobile robots. Artificial Intelligence., 2001, vol. 128, no. 1-2, pp. 99–141. doi: 10.1016/S0004-3702(01)00069-8
3. Dellaert F. Using the Condensation algorithm for robust, vision-based mobile robot localization. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 1999, pp. 10–12. doi: 10.1109/CVPR.1999.784976
4. Brubaker M.A., Geiger A., Urtasun R. Lost! Leveraging the crowd for probabilistic visual self-localization. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 2013, pp. 10–16. doi: 10.1109/CVPR.2013.393
5. ChuH., Kim D.K., Chen T. You are here: mimicking the human thinking process in reading floor-plans. Proc. IEEE Int. Conf. on Computer Vision, 2015. doi: 10.1109/ICCV.2015.255
6. Briechle K., Hanebeck U.D. Localization of a mobile robot usingrelative bearing measurements. IEEE Transactions on Robotics and Automation, 2002, vol. 20, no. 1, pp. 36–44. doi: 10.1109/TRA.2003.820933
7. Thrun S. Probabilistic robotics. Communications of the ACM, 2002, vol. 45, no. 3. doi: 10.1145/504729.504754
8. Badrinarayanan V. SegNet: a deep convolutional encoder-decoder architecture for image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, vol. 39, no. 13, pp. 2481–2495. doi: 10.1109/TPAMI.2016.2644615
9. Shelhamer E., Long J., Darrell T. Fully convolutional networks for semantic segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, vol. 39, no. 13, pp. 640–651. doi: 10.1109/TPAMI.2016.2572683
10. Tateno K., Tombari F., Laina I., Navab N. CNN-SLAM: Real-time dense monocular SLAM with learned depth prediction. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 2017. doi: 10.1109/cvpr.2017.695
11. LainaI., Rupprecht C., Belagiannis V. Deeper depth prediction with fully convolutional residual networks. Proc. 4^th Int. Conf. on 3D Vision, 2016. doi: 10.1109/3dv.2016.32
12. Kendall A., Badrinarayanan V., Cipolla R. Bayesian SegNet: model uncertainty in deep convolutional encoder-decoder architectures for scene understanding. Proceedings of the British Machine Vision Conference, 2017. doi: 10.5244/c.31.57 
13. Xiao J., Owens An., Torralba A. SUN3D: A database of big spaces reconstructed using SfM and object labels. Proc. IEEE Int. Conf. on Computer Vision, 2013. doi: 10.1109/ICCV.2013.458
14. Blanco J.L. Optimal filtering for non-parametric observation models: applications to localization and SLAM. The International Journal of Robotics Research, 2010, vol. 29, no. 14, pp. 1726–1742. doi: 10.1177/0278364910364165
15. Sturm J., Engelhard N., Endres F., Burgard W., Cremers D. A benchmark for the evaluation of RGB-D SLAM systems. IEEE/ RSJ Int. Conf. on Intelligent Robots and Systems, 2012. doi: 10.1109/iros.2012.6385773 
16. Labbe M., Michaud F. Online global loop closure detection for large-scale multi-session graph-based SLAM. IEEE/ RSJ Int. Conf. on Intelligent Robots and Systems, 2014. doi: 10.1109/iros.2014.6942926
17. Horn B.K.P. Closed-form solution of absolute orientation using unit quaternions. JOSA A, 1987, vol. 4, no. 4, p. 629. doi: 10.1364/JOSAA.4.000629

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License