【摘要】 针对自适应蒙特卡洛定位(AMCL)算法全局重定位恢复时间长和移动机器人单一传感器定位的精度低等问题，提出了基于改进AMCL算法的多传感器融合定位的方法。该方法首先使用扩展卡尔曼滤波(EKF)对里程计和惯性测量单元(IMU)的传感器数据进行融合。其次，将遗传算法的DNA交叉思想加入AMCL算法的粒子滤波模块中。最后，利用二维激光雷达的点云数据修正位姿，完成移动机器人的自主定位。室内场景实验结果表明：该方法能够满足移动机器人在室内已知环境中的自主定位需求，绝对定位误差控制在2 cm左右，并且定位精度比传统AMCL方法提升了16.5%,比结合DNA交叉思想的AMCL方法提升了14.6%;比基于EKF数据的AMCL方法提升了9.5%;当机器人位置发生跳变时，重定位恢复时间较3种方法分别提升了89.3%,31.7%,24.2%。更多还原

【Abstract】 Aiming at the problems of long recovery time of global relocation in the adaptive Monte Carlo localization(AMCL)algorithm and low localization precision of mobile robot with a single sensor, a multi-sensor fusion localization method based on the improved AMCL algorithm is proposed.Firstly, the extended Kalman filter(EKF)is used to fuse the sensor data of odometry and the inertial measurement unit(IMU).Secondly, the DNA crossover idea of genetic algorithm is added to particle filter module of the AMCL algorithm.Finally, point cloud data of 2D LiDAR is used to correct the pose to complete autonomous positioning of mobile robot.The experimental results of indoor scenes show that the proposed method can meet the autonomous localization requirements of mobile robots in the known indoor environment, and the absolute localization error is controlled at about 2 cm.The localization precision is improved by 16.5 % compared with the traditional AMCL method and by 14.6 % compared with the AMCL method combined with DNA crossover idea.It is improved by 9.5 % compared with AMCL method based on EKF data.When the robot position occurs jumps, the relocalization recovery time is reduced by 89.3 %,31.7 % and 24.2 % compared with the three methods, respectively.更多还原