基于ICP匹配和贝叶斯逆传感器模型的地图构建  

Map construction based on ICP matching and Bayesian inverse sensor model

作　　者：王立玲[1,2] 苏华强 马东 WANG Liling;SU Huaqiang;MA Dong(College of Electronic and Information Engineering,Hebei University,Baoding Hebei 071002,China;Hebei Key Laboratory of Digital Medical Engineering,Baoding Hebei 071002,China)

机构地区：[1]河北大学电子信息工程学院,河北保定071002 [2]河北省数字医疗工程重点实验室,河北保定071002

出　　处：《激光杂志》2020年第12期50-56,共7页Laser Journal

基　　金：国家自然科学基金(No.61703133);国家重点研发计划(No.2017YFB1401200)。

摘　　要：针对传统算法中存在栅格地图的熵值较高和误差较大的问题,提出了一种基于逆传感器栅格地图的改进算法。该算法利用网格单元被占用时,被该网格单元遮挡的其他网格单元的占用概率与前向传感器模型无关的原理,可获得激光雷达传感器所覆盖区域内所有占用网格的后验概率。在贝叶斯框架下,将先验的占用概率和激光雷达测量信息相结合,以此降低栅格地图的熵值;同时,通过ICP激光雷达数据匹配算法来获得匹配结果,以此来代替在栅格地图构建的过程中误差较大的里程计度数,从而提高栅格地图的精度。经仿真实验表明,所提出的改进算法相对于传统对数框架的栅格地图算法而言,地图的熵值更低且熵值的平均峰值收敛速度提高了39.70%。An improved algorithm based on inverse sensor grid map is proposed to solve the problems of high entropy and large error in traditional algorithm.This algorithm uses the principle that when the grid cell is occupied,the occupancy probability of other grid cells blocked by the grid cell is independent of the forward sensor model,and the posterior probability of all occupied grids in the area covered by lidar sensor can be obtained.Under Bayesian framework,the prior occupancy probability and lidar measurement information are combined to reduce the entropy value of grid map.At the same time,ICP lidar data matching algorithm is used to obtain matching results,which can replace odometer degree with large error in the process of grid map construction,so as to improve the accuracy of grid map.Simulation results show that the entropy of the improved algorithm is lower than that of traditional logarithmic framework grid map algorithm,and the average peak convergence rate of entropy is increased by 39.70%.

关 键 词：栅格地图 贝叶斯框架 ICP算法 逆传感器模型 

分 类 号：TN958.98[电子电信—信号与信息处理]