A Robot Navigation Algorithm Based on Sparse Landmarks

ALV(average landmark vector) algorithm occupies an important position in the field of visual homing, it has many outstanding advantages such as simple model and less storage space, etc. ALV algorithm uses all the feature points within the whole image region as landmarks, which makes the calculation amount larger and at the same time is unable to ensure the correspondence of each landmark. Because of the problems´ being, an improved ALV algorithm which is based on the features of horizon ring area is proposed. The improved ALV algorithm just extracts SIFT(Scale Invariant Feature Transform) features as landmarks within the ring area, as a result of which it effectively ensures the correspondence of landmarks and decreases their quantity as well. Experimented in several test scenes, this algorithm effectively improves homing accuracy and reduces the calculation amount.