Transporting shipments reliably and quickly from the shipper to the consignee is the core purpose of the logistics business. Every dispatch order contains key information about the shipper, consignee, freight type and quantity, and specific delivery requirements. All this information, along with other parameters, forms the basis of the work of a regional LTL carrier’s dispatcher who assigns freight shipments to individual vehicles of the fleet, drivers and daily routes.
The more shipments a carrier has to transport, the larger the volume of information that must be processed, and the more complex the dispatcher’s job. Dispatchers use transport management software (TMS) which can perform a limited number of tasks more or less automatically. Provided that the available data is accurate and complete, the individual steps of the workflow can be automated easily, saving time and costs.
However, experience shows that in many cases the data furnished for a shipment are vague or incomplete. Here are two typical examples:
- The dimensional and weight information for an item is either missing or inaccurate. This means that the dispatcher cannot be sure that the surface area available on the bed of a specific lorry will be sufficient, or that the permissible maximum weight rating will not be exceeded.
- Furthermore, not every vehicle can be used for every customer: Sometimes the loading dock may be too small for a 12-tonne lorry, or a specific driver may be prohibited from delivering to a specific consignee.
There are often additional planning criteria that are nowhere to be found in the stored or printed shipment or fleet information but need to be accounted for nevertheless. Dispatchers must remember this information at all times, which is tedious and an additional stress factor.
Machine Learning (ML) and Artificial Intelligence (AI) can help dispatchers handle the multitude of factors which must be considered for effective planning: Integrating ML and AI functionalities into an existing Transport Management System will create a ‘learning’ software environment capable of recognising patterns, deducing rules and applying them in certain situations automatically. For example, if a dispatcher regularly assigns freight units of a certain type to a specific vehicle although the specified dimensions would suggest that they are actually too large, the software will conclude that this is a pattern (i.e., ‘this item for that consignee is in reality smaller than indicated in the documentation and will fit into the space available on vehicle X’). The software will then automatically assign these freight units accordingly in future.
Another example: The software detects that a specific driver never delivers to a specific destination. It concludes that this is a rule to be applied consistently. As the software makes these kinds of observations and extracts rules, it will progressively make the dispatcher’s life easier. Over time, it relieves the dispatcher of many time-consuming detail steps, freeing him up to focus on utilising the fleet and the drivers efficiently, optimising routes and reducing costs.