3 Ways Machine Learning Is Transforming the Logistics Industry

Logistics companies are using artificial intelligence and machine learning to ensure the best results to keep productivity at its highest level, make better business decisions, and keep up with the competition. The importance of AI in this industry is huge. It is estimated that in the next 20 years, companies will derive between $1.3 trillion and $2 trillion a year in economic value thanks to this advanced technology in manufacturing and global supply chains. 

If you’re still wondering how AI and machine learning can help your business, take a look at some exciting use cases and decide if this is the solution for you.

1. AI-Based Route Planning Software

​​Choosing the optimum route, planning breaks for drivers, and avoiding the most crowded and dangerous paths are just some of the many challenges that are part of daily work in the logistics industry. 

According to Goldman Sachs, when we are talking about delivering just 25 packages, possible routes reach around 15 trillions of trillions. And this is where machine learning comes to the rescue. ML-based route planning software can analyze all of the options to choose the optimal solution in terms of costs, applicable deadlines, and unexpected road events requiring immediate decisions.

Based on big data sets provided to the system, such as information about fuel efficiency, possible traffic accidents or obstructions, vehicle size, and other drivers’ work schedules, real-time route optimization algorithms determine the best route for drivers. They are cloud-based, so all information is provided in real-time and can be accessed by dispatchers, drivers, managers, and other employees, such as account managers, to keep customers informed about the expected delivery time.

Based on machine learning, route optimization software can bring many benefits to your business, such as:

• Improved customer experience: With more accurate delivery time estimates, customers will be more satisfied with your service and more likely to give you positive feedback. What’s more, you can also introduce notifications about an upcoming delivery via email or SMS. 
• Cost savings: One of the key benefits of machine learning is usually the savings in time and money. This is true here, as route optimization systems monitor fuel consumption and suggest the most cost-effective routes. 
• Monitor driver performance: A cloud system based on machine learning helps you to supervise your employees’ work and make sure they are performing their duties reliably. You can also make sure they are following the road rules and their work schedule. Moreover, being aware that managers have access to this information can increase employee efficiency and productivity.
• KPI tracking: With insight into key information such as travel time, fuel costs, and employee productivity, you can better monitor your company’s performance and react faster if any element needs improvement.

One real-life example where algorithmic route optimization improved revenue in the logistics industry is this case study from McKinsey. Their client was an Asian logistics company that asked the tech company to solve their problem with matching fleet supply and routes to customer requirements.

How did they achieve this?

First, McKinsey’s team collected all essential data about their processes to find any issues to improve. They analyzed vital information such as customer locations, hub locations, and fleet resources. ​​This information allowed them to build a route optimization model that generates customized schedules for all vehicles. With this solution, they were able to improve management in many areas, taking into account factors such as:

• Type of the vehicle
• Utilization cost
• Maximum load-outs
• Travel time

What was behind their success?

It was both the experience and the cutting-edge Machine Learning algorithms they used to build this solution. For instance, they used the Network Optimization Algorithm (NOAH) model to build visual guides in daily maps of the routes. Additionally, they provided a mobile app showing real-time data, making for easier work for both dispatchers and drivers.

As a result, their solution reduced costs by 3.6% and increased efficiencies of the line-haul network, which led to a 16% profit increase.

2. Chatbots in Logistics

Did you know that as many as 97% of people say bad customer service has an impact on their buying intentions? However, another resource says that 36% of customers are still frustrated by the failure of companies to respond to their simple questions. 

This data shows the importance of having a chatbot to respond to customers immediately to save time and improve customer experience. Virtual assistants use natural language processing to talk to people on a chat, usually right on the company homepage. They are built with algorithms that can recognize the question asked and then match the answer to it. Suppose a user asks an incomprehensible question for which there is no answer in the database. In that case, the chatbot tries to match one of the “fallback” answers or learn new patterns from the customer to use this information the next time a similar question is asked. 

A chatbot has a certain amount of knowledge about a company and its products or services. It can use its databases or draw information from external sources. The virtual advisor answers questions and conducts the conversation itself, directing the conversation to topics related to the company’s activities or suggesting a visit to a related page.

5 Key Benefits of Chatbots

Still not sure that chatbots are a good solution for your business? Just take a look at five key benefits of implementing them in a logistics company.

1. Immediate Responses 24/7/365

In logistics companies, customer contact is crucial. For example, DHL offers three different contact forms:

• Email to customer service
• Telephone contact
• 24/7 chatbot

The chatbot allows customers to get instant information about shipping status, pricing, the expected delivery time of a package, and more.

Why is it important?

Today, 77% of people expect to get immediate responses from the online chat at any time of the day or night. Chatbots can work all the time, even when your employees are not working (plus, they will never be tired). 

Implementing a chatbot that is always available significantly improves the user experience. For example, with the Helmi chatbot created by GetJenny, The Foundation for Student Housing in the Helsinki Region noted an increase in its overall customer service satisfaction score from 4.11 to 4.26. 

2. Better Website Navigation

Did you know that 34% of customers are frustrated by difficult site navigation? 

Chatbots can solve this problem by helping the visitors navigate the site and quickly find the information they are interested in. They help you with creating a positive brand image and personalized customer experience. So if you care about building satisfaction and brand loyalty among your customers, a chatbot can be an excellent first step. 

An interesting example of a chatbot that helps you find all the information about a product is the chatbot Alex, available on the Intellexer Summarizer website. When you ask him a question, you will receive a message with a link to a page where you can find information of interest.

To create such a bot, you don’t need to provide and extract a lot of data. You just need to process the content of the website to provide it in an appropriate form. Then, you separate the information about the content of the page and the data to create a logical flow of the conversation. Moreover, chatbots are constantly learning, so the more questions they receive, the more accurate their answers will become. Often, this type of chatbot is the first AI solution that companies opt for.

3. Delivery Assistance

Virtual assistants can be the first contact with customers and receive delivery requests from them. Like other AI solutions, they can relieve your employees of many repetitive tasks, such as collecting order information. What’s more, they can also instantly execute delivery-related customer requests, such as sending an invoice for an order or informing about the delivery status.

4. Comprehensive Employee Support

Chatbots can help your employees in many ways, from paperwork to placing orders to processing payments. They can receive or fill documents such as invoices or payment requests, and many more. And when machines need human assistance, they send a message to human workers to make the right next step. 

According to Bas Vogels, supervisor and trainer of the DHL customer service team: “Employees have much more time to sort out complex customer questions and prevent escalations. The employee satisfaction rate has also increased enormously.”

5. Real-Time Shipment Tracking

In logistics, delivery time and real-time information about the status of an order are crucial. Chatbots will make sure your customers don’t have to wait for a response. A real-life example of this solution is the case study from RoboRobo. They created a bot for RPL that informs customers about their order’s status. The chatbot allows RPL’s customers to monitor the location of their package and find out when it will be delivered.

Chatbots can be used in many places, not only on a website. More and more companies are opting for chatbots available on Facebook, Skype, WhatsApp, and other channels.

3. Solving Picker Routing and Batching Problems in Warehouse Operations

Another task that artificial intelligence fulfills in logistics is to develop the most efficient methods for the flow of goods both in the warehouse and in the distribution phase.

AI-based warehouse management systems can record all activities and processes taking place in the warehouse. The software analyzes the historical data collected and uses it to plan how the equipment used (robots and both automatic and semi-automatic systems) will handle the loads. Especially helpful here can be deep learning, predictive analytics, computer vision, and product recognition software that can help recognize objects in the warehouse and make extended forecasting of what actions will be needed.

One of the main goals of machine learning algorithms is to help people with monotonous but hard tasks. In the logistics and manufacturing industry one of these tasks is the picker routing, which machines can also support. 

An exciting example of this is the solution created by Nvidia for Zalando, an e-commerce giant, which has thousands of new orders every hour. Their AI-based solution allowed to solve two problems.

​​1. Reducing Picker Routing Time

They prepared a solution allowing warehouse control with a “rope ladder” layout (which means that all products are stored in shelves placed in several rows with aisles). Given that a worker needs to retrieve products located in different warehouse parts, the system suggests the shortest possible route across the warehouse that allows picking all the items needed. 

The Nvidia’s developers created the OCaPi (Optimal Cart Pick) algorithm that finds the optimal pick tour for the worker and even for the movements of the worker’s cart. It allowed Zalando’s workers to quit using the S-shape routing heuristic and plan a more optimal route.

2. Solving the Batching Problem

At Zalando, all orders have to be assigned to a pick list. When the list is complete, the products are packaged for the customer.

The Nvidia developers tried to make a solution that allows achieving the sum of the travel times for all picklists as small as possible, assuming that a worker can fit only 10 items into the cart. They analyzed OCaPi pick tours for ten orders of two things to find the most efficient splits of orders into pick lists.

What Technologies Can Reduce These Problems?

A key technology used in these projects is the OCaPi algorithm — a highly nonlinear function that allowed developers to calculate the travel time, considering different pick-up positions. This solution showed them that travel mainly depends on the time spent picking an item from the back corner, placed far from all other products. 

To make OCaPi travel time estimation even faster, they used Caffe neural network framework and NVIDIA’s cuDNN convolutional neural network library. It allowed them to train four models in parallel to find a very accurate neural network architecture. As a result, their system allowed the company to decrease the travel time per item picked by about 11%.

Such machine learning-based solutions allow companies to:

• Increase productivity
• Speed up order picking times, resulting in increased consumer satisfaction
• Increase the satisfaction of employees whose work is supported by intelligent solutions
• Improve employee daily workflow
• Eliminate human error since route calculation is quicker and more accurate than if a human did it.