The invention provides a method for collaborative mapping and locating of multiple robots for a large-scale environment. The method comprises a single-robot laser SLAM algorithm based on a visual detection closed loop, a multi-robot pose constraint estimation algorithm and a multi-robot map fusion algorithm, wherein according to the single-robot laser SLAM algorithm based on the visual detection closed loop, a visual sensor is adopted for assisting a laser sensor in achieving the SLAM algorithm with the more stable roughness. Simultaneous locating and mapping of the multiple robots are achieved through the laser sensor and the visual sensor. The closed loop is detected by obtaining the visual characteristic of the roughness through a camera, and the problem about closed loop detection caused by the robot motion accelerative error is solved effectively; meanwhile through a multi-robot system, simultaneous locating and mapping in the large regional environment are completed efficiently, and the defect that the efficiency is low by means of a single robot is overcome. By the adoption of the method, the precise robot location and map creation of the environment are achieved in the large-scale environment, and the method is also suitable for small-scale environments.