The invention relates to an unmanned ship inland river obstacle sensing method based on a
laser radar. The method comprises the steps that 1, constructing a polygonal geographic electronic fence area of a task
water area online in real time by taking the position of a ship as a reference point; 2, carrying out
distortion correction on the 360-degree
point cloud generated by the
laser radar, and combining the processed
point cloud information with the position of the ship to carry out filtering according to the geo-fencing area in the step 1; 3, carrying out horizontal plane grid projection dimension reduction
processing on the corrected
point cloud, carrying out statistics on multi-dimensional feature information of the point cloud belonging to each grid, and calculating a target confidence coefficient of each grid according to a feature value; 4, performing
noise removal, target grid segmentation and clustering
processing on the grid map, extracting position and scale information of each target, and obtaining a target two-dimensional rectangular bounding box; and 5, extracting position coordinates of an obstacle target bounding box and bounding box length-
width ratio information according to the front and rear frame grid maps of the
laser radar, performing front and rear frame clustering target association matching, and extracting target position speed information. According to the invention, accurate acquisition of the position and size of the target is realized.