In a collaborative project with Adiona, researchers from the University of New South Wales have published new findings on optimizing last-mile delivery routes. Funded by the Australian Research Council, the study focused on the gap between mathematically optimized routes and what “optimization” means in the real-world. The research takes into account the multiple competing priorities that fleet owners and operators have to consider when optimizing their delivery routes.
Using 800+ real-world scenarios from Adiona data, researchers developed a data-driven predictive-prescriptive (DDPP) framework that compares the optimized route against the route that was actually driven. Using machine learning, the framework learns the user’s preferences by analyzing the minor adjustments made by users and predictively applies these to future routes.
The multiple priorities factored into the optimization included, in order of priority as set by the real-world decision makers:
- Minimizing the total travel distance
- Minimizing the number of vehicles required to complete the deliveries
- Minimizing the workload imbalance, i.e. the difference between the longest and shortest routes in terms of customer count
Interested in reading the study for yourself? Access the peer reviewed paper.
Practical application of last-mile delivery route optimization research
Routes that were optimized using the outputs of the research were more closely aligned with decision-maker preferences, and had less edits to the routes. These routes maintained efficiency even while considering multiple priorities.
Reducing the time spent reviewing and tweaking routes gives time back to dispatch teams. This could mean shorter turnaround times from receiving orders to getting delivery vehicles out, to fitting more routes into a day and allowing drivers to pick up extra routes, or re-allocating resources to other optimization projects.
Bridging theoretical route optimization with practical application: Exciting new Adiona AI features coming soon
The collaboration between Adiona, UNSW, and the Australian Research Council means this research will be implemented into Adiona’s route optimization platform, FlexOps.
Coming soon to Adiona’s platform is an inbuilt AI feature that analyzes your routes and the edits you’ve made to automatically apply your preferences to future route optimizations, without any additional steps. Adiona’s routing is already considered extremely accurate by our customers, and is set to become even more accurate, taking into account your specific definition of “optimized” and adding a new layer to the ways AI is transforming the logistics industry.
Try Adiona’s route optimization, with full access and completely free for 14 days. Create an account here. Want to see some tutorials? Check out our interactive library of scenarios and walk throughs.