A robot navigation method based on laser and 2D code

In this paper, we present a method of sensor data fusion to solve some problems in the robot navigation process. As we all known, 2D range laser has the advantage of high precision, long distance and obstacle avoidance. It is very common to solve the problem of Simultaneous Localization and Mapping (SLAM) during navigation by using lasers. However, if we use the laser-only method to build our environment map, some wrong matching scan of lasers may be occurred in the scenarios with symmetric structure like offices or geometrical simplicity like long and narrow corridors. Especially, when we detect a wrong loop closure with a graph-based SLAM algorithm, the map we build may have a destructive result. In this paper, a type of 2D code, which can provide unique identification of a place, is used to overcome this problem. On the other hand, 2D codes can not only provide a robot with initial pose estimation, but also adjust the robot pose for largely cumulative errors of other sensors. In our experiments, we tested the localization accuracy of 2D code, and then built the 2D occupancy grid map (including 2D codes) in two real environments on two different robots respectively. The results show the effectiveness of our modified based-graph SLAM algorithm.