| 基于差分进化的Unscented FastSLAM2.0算法 |
| |
| 引用本文： | 吴迎国,于春梅.基于差分进化的Unscented FastSLAM2.0算法[J].井冈山大学学报（自然科学版）,2016(6):48-54. |
| |
| 作者姓名： | 吴迎国 于春梅 |
| |
| 作者单位： | 西南科技大学信息工程学院,, 四川, 绵阳 621010;西南科技大学特殊环境机器人技术四川省重点实验室, 四川, 绵阳 621010,西南科技大学信息工程学院,, 四川, 绵阳 621010;西南科技大学特殊环境机器人技术四川省重点实验室, 四川, 绵阳 621010 |
| |
| 基金项目： | 特殊环境机器人技术四川省重点实验室开放基金项目（13zxtk06） |
| |
| 摘 要： | 针对Fast SLAM2.0算法中重采样过程带来的"粒子耗尽"问题,将差分进化引入进来,提出一种基于差分进化的无迹Fast SLAM2.0算法。首先采用unscented粒子滤波器估计机器人的路径后验概率,然后采用扩展卡尔曼滤波器对环境路标进行估计和更新,最后引入改进的差分进化算法代替重采样过程来优化粒子。仿真实验表明,与Fast SLAM2.0算法相比,该方法提高了机器人在路径估计和路标估计上的精度,验证了算法的有效性。
|
| 关 键 词： | 同时定位与地图构建 移动机器人 粒子滤波器 unscented卡尔曼滤波器 差分进化算法 |
| 收稿时间： | 2016/7/10 0:00:00 |
| 修稿时间： | 2016/9/23 0:00:00 |
FASTSLAM2.0 ALGORITHM BASED ON UNSCENTED DIFFERENTIAL EVOLUTION ALGORITHM |
| |
| WU Ying-guo and YU Chun-mei.FASTSLAM2.0 ALGORITHM BASED ON UNSCENTED DIFFERENTIAL EVOLUTION ALGORITHM[J].Journal of Jinggangshan University(Natural Sciences Edition),2016(6):48-54. |
| |
| Authors: | WU Ying-guo and YU Chun-mei |
| |
| Institution: | School of Information Engineering, Southwest University of Science and Technology, Mianyang, sichuang 621010, China;Key Lab. of Robot Technology Used for Special Environment of Sichuan Province, Southwest University of Science and Technology, Mianyang, Sichuang 621010, China and School of Information Engineering, Southwest University of Science and Technology, Mianyang, sichuang 621010, China;Key Lab. of Robot Technology Used for Special Environment of Sichuan Province, Southwest University of Science and Technology, Mianyang, Sichuang 621010, China |
| |
| Abstract: | Resampling process often leads to the sample impoverishment problem in FastSLAM2.0. In order to solve the problem, a SLAM method based on differential evolution (DE) is presented by introducing DE''s idea into the unscented FastSLAM2.0. Firstly, it estimates the robot path with unscented particle filter, and uses extended Kalman filter to estimate and update the map. Then using the improved differential evolution algorithm replaces resampling process. Simulation results show that:compared with FastSLAM2.0, the presented method improves the accuracy of robot path and the landmark positions estimation. Results show the validity of the proposed algorithm. |
| |
| Keywords: | simultaneous localization and map building (SLAM) mobile robot particle filter unscented Kalman filter differential evolution algorithm |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《井冈山大学学报（自然科学版）》浏览原始摘要信息 |
| 点击此处可从《井冈山大学学报（自然科学版）》下载免费的PDF全文 |
|
