Optimization Design and Experimental Study of Gmapping Algorithm
Zhifeng Su, Jiehua Zhou, Jiyang Dai, Yongguo Zhu
Abstract
This paper aims at the issue of simultaneous localization and mapping of indoor mobile robot. In order to solve the defect of large computation and poor real-time performance of traditional particle filter algorithm, the Gmapping algorithm based on Rao-Blackwellized particle filter introduces the improved proposal distribution with high-precision lidar data and adaptive resampling measure to construct two dimensional grid map. With the help of the robot operating system, the map construction of Gmapping algorithm under different parameters is realized in two dimensional robot simulator. The performance of optimized Gmapping algorithm is verified on the platform of differential mobile robot. Finally, the fusion of multi-sensor information and optimization design of Gmapping algorithm are realized by calling the /imu/calibrate_imu service, updated and calibrated inertial measurement units data, and extended Kalman filter package. The simulation and experimental results show that the robot system can construct high precision environment map under the optimal parameter configuration.