Indoor mapping using gmapping on embedded system

Qinjie Lin, Zhaowu Ke, Sheng Bi, Sirui Xu, Yuhong Liang, Fating Hong, Liqian Feng

Abstract: In recent years, the core technology enable the rapid rise of autonomous vehicles like the solving the Simultaneous Localization And Mapping (SLAM) problem on the embedded system. And the gmapping package on the ROS platform providing laser-based SLAM is widely used and studied. However, implementing it on embedded system imposes two challenges to us: running ROS requires large CPU consumption and the algorithm of SLAM is computationally intensive. In this paper, we present a system implementation of the gmapping on an ARM based embedded hardware not installing ROS and we also improve the architecture of the mapping system by parallel execution of mapping and estimating pose. The mapping process is running on the embedded system and it can create a 2-D occupancy map from laser and pose data collected by a mobile robot with low CPU load and time consumption. Experimental results carried out with mobile robots in indoor environments illustrate the accuracy of our implementation and the low consumption of time and CPU load of the mapping system.