In this project we investigated several different schemes of barcodes. These barcodes are used in our vision-guided mobile robot system to identify and distinguish visual landmarks. We designed and implemented a barcode recognition algorithm that uses a least-square fit to evaluate a set of interpretation hypotheses of the barcode to be decoded. The cyclic, scale-invariant (shrinking) barcode design allows for partial occlusion and reading from different distances. Limitation of repeated bits, as one of the most general ways to curtail the decoding ambiguity due to lacking scale information, is applied to all our barcode schemes.