Abstract: A fusion positioning method, comprising: sampling an environmental object by means of a two-dimensional laser radar carried by a robot, so as to acquire radar point cloud-related data of the environmental object; performing adjacent matching on the position Pi of a landmark Li in the environmental object on a road sign map and the landmark map, so as to acquire the position Pj of a landmark Lj nearest to the position Pi on the landmark map; performing scan matching on a two-dimensional grid map by using point cloud data, from which distortion is removed, of the environmental object, so as to acquire a global pose of the robot in the two-dimensional grid map, and taking same as a global pose observation value G pose of the robot; and according the position Pi of the landmark Li on the landmark map, the position Pj of the landmark Lj on the landmark map, and an estimated pose and the global pose observation value G pose of the robot, performing an iterative optimization on the pose of the robot until the error