Currently, robot applications in warehousing are limited to very simple pallet loading where only one type of carton is palletized at any one time. This research investigates the situation where many cartons of various sizes must be placed on one pallet. This arises in retail businesses such as grocery or merchandise distributions;A mixed 0/1 integer programming model has been developed to solve for three-dimensional optimal pallet patterns. Since the pallet packing problem has been classified as NP-hard, a heuristic dynamic programming algorithm has also been developed. Only \u22good\u22 solution may be obtained; however, less computation time is required when compared with the mixed 0/1 model. The determined pallet pattern which maximizes the utilization of the pallet cube is then used as the input data to a robot control program for automatic palletizing;A rhino XR-2 robot is employed to investigate automatic palletizing operations. A coordinate transformation program which allows the conversion of the Cartesian coordinate to the robot\u27s joint coordinate has been completed. Therefore, only the x, y, z coordinate values of box\u27s placement location are required to the robot control program. Two efficient palletizing methods are developed. One is dynamic pallet pattern, which will dynamically select a best match pattern according to incoming box sizes. The second is multi-pallet packing with turntables, which allows robot to simultaneously load two or more pallets. Two major simulation results are presented in this research. One is the simulation statistic for multi-pallet packing. Simultaneous loading of 1, 2, 3 and 4 pallets is investigated. The other simulation is to determine the length of look-ahead box queue on the conveyor in order to dynamically select a best match pallet pattern. A miniature robotic palletizing cell is employed to collect the palletizing statistics for evaluating the performance of the developed system and determining design alternatives.
展开▼