A major challenge in Logistics is dealing with uncertainty between the physical operations and information flows. DIAL researchers have in the past developed the Intelligent Product paradigm, which enables the physical product to be cognizant of its status throughout operations by connecting it to information over its status. Ideally, the Intelligent Product should also be able to communicate with other participants of the operations and even make decisions on its own destiny.

The DASHLog project is an example of the Intelligent Product approach. The project involves a logistic and supply chain company in China. A goal of DASHLog is to design intelligent agents responsible for making one of the most valuable decisions when processing outbound orders in a warehouse: creating picklists. The strategies of grouping orders into a picklist affect how efficiently the resources (e.g., pickers, equipment etc.) would be allocated and used. The original picklists are produced by high-level agents applying predetermined strategies (e.g., based on order structure, order complexity, product popularity and product weight and volume). After that, each of the agents holds an original picklist and negotiates with other participants to exchange orders in their picklists for optimising local (i.e., time to process its own picklist) and global (i.e., overall time to process all the picklists and utilisation of resources) objectives. The final picklists are settled after the crucial improvement in this way. While pickers implement picking jobs according to the final picklists, the high-level agents track the performance of actual operations to build more effective policies for applying in future runs, thus forming a continuous reinforcement-learning feedback loop. This is a promising approach to applying the Intelligent Products paradigm in DASHLog since the decision-making becomes more dynamic and flexible, which empowers physical operations more adaptability and resilience when facing variations and disruptions.