A Mapping Method for Indoor Mobile Robot

Simultaneous location and mapping (SLAM) is a core issue in the field of mobile robots. This paper proposes an endpoint features based mapping method for an indoor mobile robot. The robot collects sensor information over some time to build a local map, and the local maps are fused to get a global map. This article defines the concept of endpoints and gives each endpoint a unique descriptor. Local endpoints and global endpoints are compared using descriptor matching and brute force matching to obtain a calibration. The local grids are merged into the global grids if the calibration is less than a threshold. Otherwise, running a pose correction for the mobile robot. Experiments show that in various indoor environments, this mapping method can obtain a grid map that is similar to the actual environment, which supports the mobile robot to complete navigation, obstacle avoidance, planning, and other works.