Transportation system is a discrete, stochastic, dynamic, multi-factors and multi-objective system in Automated Storage and Retrieval System (AS/RS). In the practical running conditions of transportation system, to deal with conflict of equipments and assignment problem of Automated Guided Vehicle (AGV), a mathematical model with multiple complex constraints is constructed to AGV dispatching optimization problem. A novel hybrid genetic algorithm has been proposed based on genetic algorithm and Lamarckian evolution algorithm and it combines them with local search technique. The method solved the multi-parameter model and the technical hard problem of searching efficacious adjacent structure, it has the ability of searching global solutions. Tests demonstrate the approach overcomes the phenomena of deadlock in transportation process. It also improves the use of resources of transportation system effectively.