This paper describes an effective method of obstacle detection by ALV (Autonomous Land Vehicle) equipped with two 2D laser range finders (LRF) installed at different locations of the ALV to obtain comprehensive information on the environment. The data processing includes two main steps: (1) data-processing of the current sample; (2) fusion of the former range data and the current one. The rough description of the ALV’s environment via the four sub-steps ( Data Filter, Obstacle Extraction, Obstacle Merging, Distinguishing Obstacle from Road-Edge) was not reliable enough for our control system. To overcome the shortcoming of the 2D LRF and the motion noise of the ALV, a Kalman filter was used to estimate the position of the obstacles; then the data of the two LRFs were collated to obtain the height and width of the obstacles. Experiment results attested the feasibility of the detection system. This paper describes an effective method of obstacle detection by ALV (Autonomous Land Vehicle) equipped with two 2D laser range finders (LRF) installed at different locations of the ALV to obtain comprehensive information on the environment. 1) data-processing of the current sample; (2) fusion of the former range data and the current one. The rough description of the ALV’s environment via the four sub-steps (Data Filter, Obstacle Extraction, Obstacle Merging, Distinguishing Obstacle from Road-Edge) was not reliable enough for our control system. To overcome the shortcoming of the 2D LRF and the motion noise of the ALV, a Kalman filter was used to estimate the position of the obstacles; then the data of the two LRFs were collated to obtain the height and width of the obstacles. Experiment results attested the feasibility of the detection system.