Collaborative planning processes span multiple planning domains and aim to create a common and mutually agreed upon plan by exchanging relevant data between planning domains. Key issues are

• sharing relevant information between companies, such as transport plans, for the purpose of flexible, situation-dependent, collaboration,
• creating trust between partners,
• sharing resource capacities, such as vehicles, storage facilities and handling equipment,
• synchronizing activities
• developing joint plans.

The latter requires planning mechanisms that strive for optimality at a network level regarding sustainability and profitability as well as for acceptability at the level of each party in the collaboration process, i.e. consistent with the company's goals to reduce costs, to increase asset utilization and revenues (transport and transshipment equipment) and to improve service.

In literature, multi-agent systems are considered to be suitable for complex logistics environments with multiple dependent decision makers, each with their own goals. Advantages are the simplicity and robustness of such systems, because very complex problems are solved by distributing them to autonomous decision makers (agents).

Although there is plenty of literature on multi-agent-systems within the area of information and communication technology, it most often deals with characterizing agents and multi-agent systems, information architectures and the advantages for information system development and software maintenance. Results on logistic planning and control mechanisms for specific applications are limited.

Within Transumo project Diploma, new scientific issues arise from the development of multi-agent based planning systems that contain the following new aspects:

• real-time, dynamic planning based on new information about orders (new orders, modifications of current orders) and resource status;
• inclusion of stochastic processes, such as variations in travel times of vehicles and handling times at terminals;
• multi-actor planning;
• dynamic pricing as an instrument for maximization of revenues and resource utilization.

The project is innovative in that multi agents systems are tested in simulations which are directly based on a real-life setting using recent and detailed real world data. Whenever possible, pilot applications are tested in the real world and experiences are fed back into the fundamental research work packages.

Participants
University Twente, Erasmus University Rotterdam, Technical University Delft, Init8, Merba, Port of Rotterdam, TNO, Ministery of Transport – DGTL

Projectmanager
University Twente: Prof.dr. J. van Hillegersberg

For more information, please contact Transumo at info@transumo.nl, +31 79 3470950