Abstract: In view of the uncertainty in landmark group based robot localization, an artificial landmark distribution method based on multi-ary M-sequence is proposed. The longest period characteristic of M-sequence is utilized to generate landmark distribution, which satisfies that all landmark groups are globally unique, with least landmark species. The difference among landmarks is expanded by reducing the landmark species, and thus the anti-interference ability of localization is improved. The efficiency of the proposed method is demonstrated by results of experiments on the substation inspection robot system and simulations under different measurement noises.