Abstract：A 2D map fusion algorithm based on covariance intersection (CI) factor graph is proposed for the problem that the covariance is unknown in the process of multi-robot pose estimation. Firstly, the transformation from the robot coordinate system to the global coordinate system is realized by the coordinate transformation matrix. Then, the local estimation information weight is obtained based on the principle of minimized non-linear performance. The local estimation information is merged by the algorithm to calculate the pose and cross-covariance of the fusion point. Finally, the covariance common formula is used to calculate the probability constraint (covariance) of the fusion nodes to the next-level variable nodes, thereby the factor graph fusion is completed. The experimental results show that the algorithm is feasible.
[1] 蔡自兴.21世纪机器人技术的发展趋势[J].南京化工大学学报:自然科学版,2000,22(4):73-78.Cai Z X. The development trend of robot technology in the 21st century[J]. Journal of Nanjing University of Chemical Technology:Natural Science Edition, 2000, 22(4):73-78.
[2] 田国会.家庭服务机器人研究前景广阔[J].国际学术动态,2007(1):28-29.Tian G H. The prospects of the family service robot research[J]. International Academic Dynamics, 2007(1):28-29.
[3] Howard A. Multi-robot simultaneous localization and mapping using particle filters[J]. International Journal of Robotics Research, 2006, 25(12):1243-1256.
[4] Lu F, Milios E. Globally consistent range scan alignment for environment mapping[J]. Autonomous Robots, 1997, 4(4):333-349.
[5] Olson E, Leonard J, Teller S. Fast iterative optimization of pose graphs with poor initial estimates[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2006:2262-2269.
[6] Frese U, Schroder L. Closing a million-landmarks loop[C]//IEEE/RSJ International Conference on Intelligent Robots and System. Piscataway, USA:IEEE, 2006:5032-5039.
[7] Grisetti G, Stachniss C, Burgard W. Non-linear constraint network optimization for efficient map learning[J]. IEEE Transactions on Intelligent Transportation Systems, 2009, 10(3):428-439.
[8] Kaess M, Ranganathan A, Dellaert F. iSAM:Incremental smoothing and mapping[J]. IEEE Transactions on Robotics, 2008, 24(6):365-378.
[9] Kaess M, Johannsson H, Roberts R, et al. iSAM2:Incremental smoothing and mapping using the Bayes tree[J]. International Journal of Robotics Research, 2012, 31(2):216-235.
[10] Kummerle R, Grisetti G, Strasdat H, et al. g2o:A general framework for graph optimization[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2011:3607-3613.
[11] 查尤平.平滑型SLAM中数据关联的鲁棒性研究[D].北京:北京交通大学,2016.Zha Y P. Robustness of data association in smooth SLAM[D]. Beijing:Beijing Jiaotong University, 2016.
[12] Thrun S. Probabilistic robotics[M]. Cambridge, USA:MIT Press, 2005.
[13] Montemerlo M, Thrun S, Koller D. FastSLAM:A factored solution to the simultaneous localization and mapping problem[C]//Eighteenth National Conference on Artificial Intelligence. Palo Alto, USA:American Association for Artificial Intelligence, 2002:593-598.
[14] Murphy K P. Bayesian map learning in dynamic environments[C]//Advances in Neural Information Processing Systems. Cambridge, USA:MIT Press, 1999:1015-1021.
[15] Olson E, Agarwal P. Inference on networks of mixtures for robust robot mapping[J]. International Journal of Robotics Research, 2013, 32(7):826-840.
[16] 钟进.基于路径图优化的多机器人同步定位与制图研究[D].哈尔滨:哈尔滨工业大学,2014.Zhong J. Pose graph based simultaneous localization and mapping for multiple-robot[D]. Harbin:Harbin Institute of Technology, 2014.
[17] 王硕,宋申民,史小平,等.噪声特性未知的多传感器协方差交叉融合姿态估计[J].控制与决策,2016,31(2):273-278.Wang S, Song S M, Shi X P, et al. Multi-sensor covariance intersection fusion attitude estimation with unknown noise characteristics[J]. Control and Decision, 2016, 31(2):273-278.
[18] 潘泉.多源信息融合理论及应用[M].北京:清华大学出版社,2013.Pan Q. Multi-source information fusion theory and application[M]. Beijing:Tsinghua University Press, 2013.
