CourierLeaving.jpg

Machine learning in logistics

Machine learning: It’s cool, it’s useful, and Airspace Technologies is all over it

Machine learning is a hot topic, and for a good reason. Right at the top, I’d say, machine learning has evolved to the point where it can – and should – be used to help virtually all businesses increase efficiency while reducing costs.

Screen Shot 2019-08-15 at 3.15.28 PM (2)

 

Defining Machine Learning in Logistics

In essence, machine learning involves using algorithms and statistical models to perform tasks typically only suitable for a human, without the need for explicit instructions. The machines sift through data to identify patterns, then test, and retest, and retest, and retestuntil they’ve identified the best solution to whatever you ask them.

This makes machine learning ideal for many industries, especially logistics. So why are more companies not using it? There are significant barriers to getting started. First, a company has to be committed – budget-wise and culturally – to onboard all the new technologies and people required to build these new services. As one of Airspace Technologies’ founders and as its present Chief Technology Officer, I’m proud to say that this is at the core of our company. When it comes to making a difference for our customers, we do whatever it takes.

The second reason competitors aren’t jumping into machine learning is that it requires highly skilled people, and they’re in short supply. Fortunately for us, the Googles of the world don’t have a monopoly on these people because many prefer a company that is smaller and less bureaucratic, yet still cutting-edge. That’s us.

This brings me to the third reason why competitors find it difficult to enter the world of machine learning. When we created Airspace, we knew it would be crucial to have all our information in a format that would make training these types of models possible once we have accumulated a sufficient amount of data. Many companies plan ahead years in advance for their business strategy, but not their data strategy. As a result, their data is a mess or at least not set up to their advantage. We set ourselves up from day one to be prepared to leverage machine learning as soon as we had enough data to start training the models.

 

Airspace14

 

So how does Machine Learning Help Airspace Customers?

The high-level answer is it drives faster and more accurate route selection and dispatching. These can both be highly complex undertakings. First, you have all the usual variables, which include decisions about which route, drivers, trucking company, airline, etc. Now add to this the individual preferences many shippers have regarding these variables. Some don’t want their shipment to go through Atlanta. Some don’t want to work with specific airlines. Some have had a bad experience with a particular airline in a specific airport and want to avoid it at all costs.

Sure, people can remember these things, but that means specific business logic exists in the brains of a few excellent people – and those people get sick, take a vacation and everyone eventually retires. Machine learning doesn’t get tired. That’s why it’s the perfect assistant for a talented employee. Instead of purely relying on people, we can use a machine that has been trained on the client dataset to help make decisions. It can make many decisions on its own, and it can make suggestions to people about the harder problems that it may not yet be able to address confidently. Machine learning dramatically increases the efficiency of both the systems and the employees in your business.

 

 
We enable humans to do what humans are good at, and machines to do what machines are good at.
Ryan Rusnak, Co-Founder & CTO

 

When a customer places an order at Airspace, our machine learning steps in and quickly solves for the routing, selecting the best person to pick up. When selecting the best driver, the algorithm considers obvious factors like distance to pickup, traffic, weather, and vehicle type, but also some key things that are less obvious – like if they have been to that site before, and how efficient the driver is on site. Drivers can have a wide variety of efficiencies when it comes to navigating a large surgical center. So it’s not necessarily about dispatching the closest driver, it’s about finding the best driver with the right skills and experience for this particular task.

Is machine learning required to run a logistics company in 2019? Absolutely not. Is it needed to minimize error, maximize efficiency, and provide unbeatable service? Definitely.

Watch Airspace's technology in action

SHARE THIS STORY | |

Search

Subscribe to Blog