AI Route Planning for Post

“AI algorithms can do many things the human brain alone cannot,” RouteSmart Technologies Director of Research and Development Damon Gulczynski says. “But the most powerful thinker in known existence is a human who knows how to use AI effectively.”

I love this quote from October 2023. I found it in a RouteSmart Technologies blog that in turn was based on an article published in the Parcel & Post magazine. When plying Chat GPT with various questions relating to the role of AI on route optimisation for postal delivery specifically, the best it returns is; “AI algorithms analyse various factors such as traffic patterns, weather conditions, and delivery schedules to optimize delivery routes. This reduces fuel consumption, delivery time, and operational costs”.

Cutting through the AI hype

At recent industry events I have been quizzing software developers and operational research specialists about the application of AI to the algorithms that solve the travelling sales problem and its generalisations the vehicle routing problem and the travelling purchaser problem. The responses were revealing in that AI is not being applied to the algorithms per se, more to the management of them. One leading supplier uses AI in the scheduling of the application of algorithms to solve the problem in question, swapping the algorithms in and out at various times and stages of processing in an attempt to produce an optimal outcome in as short a time as possible.

Understanding the problem

Efficient route planning is a complex problem, especially for parcel and postal delivery companies that must consider factors such as multiple delivery points, varying package sizes, traffic conditions, safety issues, and time constraints. This is complicated further with the explosion in ecommerce post Covid and the ever-increasing demand for postal operations to deliver parcels in combination with letter (mail) delivery. This is where data-driven route planning with artificial intelligence and machine learning come into play, offering the potential to transform route planning into an adaptive process.

AI, ML and predictive analytics

There is a difference between AI and ML (machine learning). Do a search with your favourite engine (there is more than one) to discover the difference. The key takeaway is that the AI or ML solution learns how certain data variables can affect the route, such as, the most common package destinations, historical delivery times, traffic patterns, weather conditions, and even the behaviour of individual delivery personnel.

By analysing historical data and leveraging machine learning algorithms, postal delivery operations can predict future demand with high accuracy.  This helps in allocating resources effectively, ensuring that the right number of vehicles and personnel are available to handle peak delivery times, thus reducing operational costs and enhancing service quality.

Data and processing power

AI and ML involves an inordinate amount of data crunching of vast data sets. Processing power is key to ensure AI and ML can help create solutions fast enough to make them operationally viable, ie, in time for the delivery team to sort, load and deliver.

RouteSmart Technologies and AI

Will AI have a meaningful role in route optimisation for postal services? Specifically, will the application of AI significantly improve the outputs of route optimisation, ie, make the route more optimal than what can be achieved now? If you cut through the current hype, I think the answer as of mid-2024 is not yet. But I think it will soon. Damon and his colleagues at RouteSmart Technologies are looking closely at AI/ML and its application to the parcel and postal sector. After all, the company has been at the forefront of operations research for the last 40 years and continues to enhance their route optimisation product sets to meet the demands of its customers. And no AI hype will deter RouteSmart Technologies from their relentless march toward route optimisation perfection.