基于位置细胞的空间表征及位置估计模型

doi:10.16183/j.cnki.jsjtu.2018.04.015

摘要/Abstract

摘要： 为开发具有生物启发式空间表征和自主定位能力的导航新方法，提出一种基于位置细胞的空间表征及位置估计模型.该模型通过径向基函数神经网络实现网格细胞到位置细胞的转换，生成自运动感知下的位置细胞.同时，通过环境感知和相似性度量生成视觉感知下的位置细胞.最后，采用信息加权的方式对前两种位置细胞进行融合，生成多信息感知下的位置细胞，以此表征已探索的空间.当运行体在已表征空间中运行时，基于重心估计原理对群体位置细胞放电活动进行处理，实现自主定位.仿真分析结果表明，所提模型能够实现已探索空间的内部表征，生成的位置细胞具有生物位置细胞的放电特性，且多信息感知下的空间表征在某一感知方式存在误差时仍表现出好的位置估计性能.

关键词: 仿生定位, 空间表征, 位置细胞, 多信息感知

Abstract: To develop a new navigation method with bio-inspired spatial representation and autonomous positioning ability, a model of spatial representation and location estimation based on place cells is presented in this paper. In the proposed model, the place cells of self-motion perception are generated through the transformation from grid cells to place cells based on radial basis function (RBF) neural network. Simultaneously, the place cells of visual perception are generated through environment perception and similar measure. Finally, the above two kinds of place cells are fused by information weighting, and the place cells of multi-information perception are generated to represent the explored space. When the vehicle runs in the represented space, it can realize autonomous positioning by processing the firing activity of population place cells based on gravity center estimation principle. Simulation results indicate the proposed model can represent the explored space, and the firing characteristic of the generated place cells is similar to that of biological place cells. Besides, when there is an error in one of the perception modes, the spatial representation of multi-information perception can also give the good location estimation performance.

Key words: bionic positioning, spatial representation, place cells, multi-information perception

中图分类号:

Q811
TP273

周阳，吴德伟. 基于位置细胞的空间表征及位置估计模型[J]. 上海交通大学学报（自然版）, 2018, 52(4): 488-494.

ZHOU Yang,WU Dewei. Spatial Representation and Location Estimation Model Based on Place Cells[J]. Journal of Shanghai Jiaotong University, 2018, 52(4): 488-494.

参考文献

［1］李伟龙, 吴德伟, 周阳, 等.基于生物位置细胞放电机理的空间位置表征方法［J］.电子与信息学报, 2016, 38(8): 2040-2046. DOI: 10.11999/jeit151331. LI Weilong, WU Dewei, ZHOU Yang, et al. A method of spatial place representation based on biological place cells firing［J］. Journal of Electronics and Information Technology, 2016, 38(8): 2040-2046. DOI: 10.11999/jeit151331. ［2］杨晶东, 杨敬辉, 蔡则苏.基于多目标优化的移动机器人避障算法［J］.上海交通大学学报, 2012, 46(2): 213-216. YANG Jingdong, YANG Jinghui, Cai Zesu. Obstacle avoidance based on multiple objective optimization for mobile robots［J］. Journal of Shanghai Jiao Tong University, 2012, 46(2): 213-216. ［3］EDVARDSEN V. Goal-directed navigation based on path integration and decoding of grid cells in an artificial neural network［J］. Natural Computing, 2016(2): 1-15. ［4］O’KEEFE J, DOSLROVSKV J. The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat［J］. Brain Research, 1971, 34(1): 171-175. DOI: 10.1016/0006-8993(71)90358-1. ［5］MAYA G S, LIORA L, YOSSI Y, et al. Spatial cognition in bats and rats: From sensory acquisition to multiscale maps and navigation［J］. Nature Reviews Neuroscience, 2015, 16(4): 94-108. DOI: 10.1038/nrn3888. ［6］MOSER E I, KROPFF E, MOSER M B. Place cells, grid cells, and the brain’s spatial representation system［J］. Annual Review of Neuroscience, 2008, 31(1): 69-89. DOI: 10.1146/annurev.neuro.31.061307.090723. ［7］SHEYNIKHOVICH D, GREZES F, KING J R, et al. Exploratory behaviour depends on multisensory integration during spatial learning［C］∥ICANN 2012, Part I, LNCS 7552. Berlin Heidelberg: Springer, 2012: 296-303. DOI: 10.1007/978-3-642-33269-2_38. ［8］BUSH D, BARRY C, BURGESS N. What do grid cells contribute to place cell firing?［J］. Cell, 2014, 37(3): 136-145. DOI: 10.1016/j.tins.2013.12.003. ［9］KESNER R P, ROLLS E T. A computational theory of hippocampal function, and tests of the theory: New developments［J］. Neuroscience and Biobehavioral Reviews, 2015, 48: 92-147. DOI: 10.1016/j.neubiorev. 2014.11.009. ［10］MUESSIG L, HAUSER J, WILLS T J, et al. A developmental switch in place cell accuracy coincides with grid cell maturation［J］. Neuron, 2015, 86(5): 1167-1173. DOI: 10.1016/j.neuron.2015.05.011. ［11］SOLSTAD T, MOSER E I, EINEVOLL G T. From grid cells to place cells: a mathematical model［J］. Hippocampus, 2006, 16(12): 1026-1031. DOI: 10.1002/hipo.20244. ［12］SI B, TREVES A. The role of competitive learning in the generation of DG fields from EC inputs［J］. Cognitive Neurodynamics, 2009, 3(2): 177-187. DOI: 10.1007/s11571-009-9079-z. ［13］FRANZIUS M, VOLLGRAF R, WISKOTT L. From grids to places［J］. Computational Neuroscience, 2007, 22(3): 297-299. DOI: 10.1007/s10827-006-0013-7. ［14］DALLEMOLE V L, ARAJO A F R. A novel topological map of place cells for autonomous robots［C］∥ICANN, part II, LNCS 6353. Berliin: Springer-Verlag Berliin Heideberg, 2010: 296-306. DOI: 10.1007/978-3-642-1582 2-3_37. ［15］JAUFFRET A, CUPERLIER N, GAUSSIER P. From grid cells and visual place cells to multimodal place cell: A new robotic architecture［J］. Frontiersin Neurorobotics, 2015, 9(1): 1-22. DOI: 10.3389/fnbot.2015.00001. ［16］ZHOU Y, WU D W. Grid-to-place cells model based on radial basis function network［J］. Electronics Letter, 2017, 53(3): 200-201. DOI: 10.1049/el.2016.1750. ［17］ZHOU Y, WU D W. A model of generating visual place cells based on environment perception and similar measure［J］. Computational Intelligence and Neuroscience, 2016(3): 1-9. DOI: 10.1155/2016/3253678.