CAJ | 학술논문

针对FastSLAM算法对传感器精度要求较高,不适用于方向性差的超声传感器问题,提出了一种基于超声概率栅格地图环境特征点提取匹配的移动机器人粒子滤波同时定位与地图创建(SLAM)算法.该算法可分解为机器人位姿估计和环境路标估计2个部分.基于蒙特卡罗定位原理利用粒子滤波算法对机器人运动轨迹进行估计;在建立全局超声概率栅格地图的基础上,利用概率栅格地图环境特征提取算法对环境路标坐标进行估计.实验证明,该算法较好地解决了超声测距传感器由于散射角大带来的特征点估计不准的问题,对环境路标和机器人轨迹的估计都比较准确.并对移动机器人累计误差进行了有效的补偿,减少了由于累积误差造成的移动机器人轨迹扭曲失真.
침대FastSLAM산법대전감기정도요구교고,불괄용우방향성차적초성전감기문제,제출료일충기우초성개솔책격지도배경특정점제취필배적이동궤기인입자려파동시정위여지도창건(SLAM)산법.해산법가분해위궤기인위자고계화배경로표고계2개부분.기우몽특잡라정위원리이용입자려파산법대궤기인운동궤적진행고계;재건립전국초성개솔책격지도적기출상,이용개솔책격지도배경특정제취산법대배경로표좌표진행고계.실험증명,해산법교호지해결료초성측거전감기유우산사각대대래적특정점고계불준적문제,대배경로표화궤기인궤적적고계도비교준학.병대이동궤기인루계오차진행료유효적보상,감소료유우루적오차조성적이동궤기인궤적뉴곡실진.
Focusing on the limited application of FastSLAM algorithm due to its high accuracy requir-ment for sensor and unfitness for ultrasonic sensor with poor directivity, a simultaneous localization and mapping (SLAM) algorithm utilizing particle filter is provided, which is based on ultrasonic probability grid map feature points extraction and matching. This algorithm consists of robot pose esti-mation and environment landmark estimation. Particle filters are applied to estimate the robot trajectory according to Monte Carlo methods. Based on global ultrasonic probabilistic grid map, the environment landmark position is observed by environment feature extraction algorithm. The effectiveness of the proposed algorithm is validated by experimental result. The estimation accuracy of feature point posi-tion as well as environment landmark are improved. The cumulative error of the robot is compensated effectively and the mobile robot trajectory distortion is reduced.