Abstract: A consensus fusion algorithm for multi-odometer data is presented to improve in-pipeline location precision of the autonomous pipeline inspection robot.Based on the concept of confidence distance measure,the confidence distance matrix and relation matrix for multi-odometer data is constructed firstly.And then line digraph is used to eliminate biggish inaccuracy or error of the measured data.At last,optimal fusion value of multi-odometer data is obtained by maximum likelihood estimate.Running on the intelligent controller of pipeline robot,the presented algorithm is realized with C programming language.In autonomous crawling experiment of the robot,validity of the localization method is proved by measuring space intervals between girth welds in simulative oil-and-gas pipeline.