Feature-to-Feature Based Laser Scan Matching for Pallet Recognition

Abstract

The pallet recognition method based on feature-to-feature laser scan matching is proposed. The laser scan data are clustered into one line segment if they belong to one object or adjacent objects; then the corner points are found in the line segments; therefore, the environment can be expressed by the line segments. The environment line segments are compared with the pallet model to recognize the pallet. The recognition ratio and velocity of the feature-to-feature based method high and fast. On our robot which configured Pentium III 400 MHz, 512 M RAM, the static recognition ratio can reach 100%; when the robot moves, the recognition ratio can reach 98%; the recognition velocity is 100 frames per second.