In recent years, to address labor shortages and improve productivity in the logistics industry, we have been developing an Autonomous Mobile Robot（AMR）that transports roll box pallets. During the transportation process, especially when placing the pallets, it is necessary to align the palettes as closely as possible to make effective use of warehouse space. A conventional method has been proposed to calculate a target position from detection results of the side of the pallet. However, there is a problem that the placement shifted due to slippage when the heavy pallet was transported. In this study, we propose a method to correct the placement using the detection results of the pallets wheels. This method classifies point clouds by each pallet. Then, it calculates the pose of pallets by fitting the pallet model to point clouds of wheels and re-calculate target position. As a result of aligning 95 kg and 563 kg pallets, we confirmed that three-sigma of mean error of horizontal interval were improved from 279.7 mm to 85.6 mm and those of vertical interval were improved from 195.2 mm to 191.5 mm.
MIYAMOTO et al. (Wed,) studied this question.