Among the things that did not change around us after our relocation to Chicago was the winter precipitation, a.k.a. snow. We had a decent winter storm this week and are bracing for another one this weekend. As I was looking outside this morning, sipping tea, I admired how beautiful and pristine the snow looked and how it made everything it covered look beautiful. It covers everything, dying or dead, frozen underneath and hides the ugly winter landscape. Looks beautiful till you have to deal with it. Get rid of it. Shovel it. 

Then, I realized it is a similar phenomenon to excess inventory. Excess inventory hides so many ugly aspects, which are a challenge for many organizations. And with this “linking thought,” my mind wandered from philosophical thought into a mathematical formulation. But this time, it was not around inventory. It was around transportation.

We moved from a rural area in Massachusetts to a city in Illinois. Large cities and towns have their own challenges. One such challenge is to optimize snow removal during and after a snowstorm. The “mathematical formulation” thought was around how technology can help cities further optimize this operation.

I observed that snow plows were tackling streets based on a particular hierarchy. So, I decided to research a bit more. What I found was common sense-driven but optimal to a certain extent. You tackle the busiest streets first, so the hierarchy is highways, then streets. Lanes, circles, and other less busy roads are the last. This approach made perfect sense. However, this one also has constraints like all the other mathematical formulation problems. There is a constraint on the number of snow plows available. So, even with the existing hierarchy, there needs to be sub-hierarchies. How do you decide which stretch of an interstate to clear first?

With my limited knowledge of public utilities operations, I assume someone manually leverages real-time weather information and then dispatches and directs plow routes based on how the storm hits certain areas. While this process works, this might not always be optimal. In my analytics career, I have worked on many transportation transformation processes where the bottleneck was the dispatch operations. Not because they were not efficient or productive. But, as humans, our ability to absorb many information sources and process them simultaneously to output an optimal result is limited.

And this is where AI can help.

We can develop an AI-paired optimization algorithm to help navigate the complexity of real-time, dynamic dispatching. These algorithms can also make a powerful impact in the corporate sector, wherever real-time dispatching is leveraged. Every input source, currently in the form of papers, phone calls, emails, etc., can be re-engineered to work seamlessly with such a solution.

In this specific example, the current location of plows, their capacities/capabilities, GIS data paired with weather data, information extracted from calls to an automated line, email extracts, and publicly available data (like a local news channel reporting on a traffic snarl on a particular street due to snow accumulation) can be processed in seconds, to ensure that despite the mayhem, the city can provide maximum coverage with the amount of resources that are available to it.

Ah- the beauty of technology! And after this thought, I was back to finishing my tea and admiring the pristine snow outside my window.