Pallet Pose Estimation with LIDAR and Vision for Autonomous Forklifts

Abstract This paper is focused on enhancing current AGV flexibility in non structured environments. It proposes a scheme to solve the problem of identifying a pallet, which pose is known with large uncertainty, from fused laser range and vision data and navigating to it by on line calculating and performing highly continuous paths for a precise target reaching. The novelty is in the combination of range and colorimetric measurements still not exploited, to our knowledge, for pallet recognition and localization.