Auction-based Multi-agent Task Allocation In Smart Logistic Center

As a fundamental problem in the multi-robot system research domain, task allocation is becoming more and more important with the increase of system size and task complexity. In logistic field, intellectualized warehouse management has become the trend of productivity development and multi-agent system is a suitable plan to implement this intelligence. Therefore, the thesis is dedicated to the research of this field.In the first place, the thesis discribes the classification and formal description of multi-agent task allocation, and summarizes the current main methods, styles and state-space of task allocation. Aiming at intentional cooperative methods, the thesis discusses some general methods to solve multi-agent task allocation problem, including Integer Programming, Combinatorial Auction, Graph-Searching, Tree-Searching method and centralized or distributed multi-round auction method. The content is about the key point of each method, such as applicable situation, complexity, etc as well as the reason of unsuitability of these methods in intellectualized warehouse management.Moreover, the thesis develops intellectualized warehouse management plan in which multi-agent system are involved. First, the background and key point of recent related research are introduced and the description and model of multi-agent task allocation in smart logistic center are given. Second, Aiming at task allocation problem of a great number of agents located in smart logistic center, a real-time assignment approach is developed based on the auction method. As the existence of asymmetrical shelf-shift tasks in the center and self-finished cost of each task, besides the correlation between every two tasks, the problem of multi-robot task assignment in logistic center is different from multi-robot routing problem. By introducing the correlation function and self-finished cost function and choosing suitable bidding strategy, this paper implements the static and dynamic task assignment. The simulation results show the applicability of the algorithm.