摘要
针对RoboCup四腿组比赛场地结构对称和特征不唯一的特点,在场地模型中对带数据校验的扩展卡尔曼滤波(EKF-V)、多假设定位(MHL)、蒙特卡洛定位(MCL)和自适应蒙特卡洛定位(A-MCL)四种算法的全局定位精度和对噪声的鲁棒性进行了仿真实验比较.实验结果表明,四种算法在噪声可估计的条件下都能达到较高的全局定位精度,而MCL和A-MCL对噪声有较高的鲁棒性,更适合应用于RoboCup四腿组比赛.
For RoboCup four-legged league field with symmetrical structure and non-unique features, the global localization accuracy and robustness against noises of four localization methods, including Extended Kalman Filter with data Validation （EKF-V）, Multiple Hypothesis Localization （MHL）, Monte Carlo Localization （MCL） and Adaptive Monte Carlo Localization （A-MCL）, are compared in a simulated f,eld model. The experimental results show that all the algorithms achieve high accuracy when the noise can be estimated. However, MCL and A-MCL are preferable when applying to RoboCup fourlegged league due to their robustness against noises.
出处
《机器人》
EI
CSCD
北大核心
2007年第2期167-170,共4页
Robot
基金
国家自然科学基金资助项目(60474024)
清华大学教育基金资助项目(202025001)
关键词
定位
ROBOCUP
卡尔曼滤波
多假设定位
蒙特卡洛定位
localization
RoboCup
Kalman filter
multiple hypothesis localization
Monte Carlo localization
