Abstract:Traditional visual navigation methods depend on stable environmental characteristics and globally consistent maps which have high requirements on environmental features and lighting conditions. These methods often face challenges such as the inability to obtain an accurate navigation position and convergence issues. To address these challenges, this paper presents a homing method for autonomous mobile robots based on ArUco marker recognition, without the need for map building. The proposed method involves recognizing the ArUco markers in the field of view of the robot, and sequentially recording the initial poses of the robot during the movement, the robot can be localized and navigated for homing through continuously choosing the earlier ArUco marker and building the corresponding local coordinate system. The method achieves the homing results of the robot for different navigation paths with static obstacles. In the complete coverage path test, the homing trajectory saves 79.16% of the coverage trajectory length. Furthermore, with the 20 times repeat stopping test, the robot achieved a 0.8cm average stop distance, which shows its precise navigation ability. The method is adaptable to different paths and it demonstrates a competence for skipping parts of the path and cutting corners to the start point.