Due to the impreciseness of any model representation and the fact that a significant amount of relevant information is not available to the computer, but only to the dispatchers who are in direct contact with drivers and shippers, model recommendations cannot always be fully implemented.
Interactive problem solving techniques can be used to deal with this issue. A problem can be posted to a planning method in order to let it generate a solution. The returned solution usually does not satisfy all real-life requirements. Dispatchers may add, modify, or remove certain constraints in the analytical model. The modified problem is again posted to the planning method, and after a solution has been found, further modifications to the model can be made. Alternatively, this interactive process can be realised simultaneously. Although such approaches can be used for static and dynamic problems, iterative decision making is much harder in dynamic planning due to the lack of time.
Humans use a form of cache memory for processing information, called working memory. This working memory must be regularly updated in order to consider changes in problem data and solution. A dynamic planning system must support dispatchers in quickly updating their working memory. Therefore, besides of providing algorithmic optimisation techniques, a dynamic planning system must also provide sophisticated graphical user interfaces (GUI) allowing dispatchers to quickly identify modified parts of the solution, and to efficiently verify feasibility and profitability of an automatically generated schedule.