© Kittipong Jirasukhanont | Dreamstime
Robots In Warehouse 60bfd5974f90f

Data Replicas, Digital Twins and AI (Oh My!)

June 8, 2021
How disruptive technology can drive innovation in the warehouse.

The warehouse of the future is a man and a dog. The man is there to feed the dog. The dog is there to make sure the man doesn’t touch anything. More frequently in supply chain circles, we hear this concept referred to as a dark warehouse. From a technologist’s perspective, all of the recent automation introduced at distribution centers makes it seem as if the movement towards a fully autonomous dark warehouse isn’t too far away.

However, have you ever talked with the folks running operations? When you have a frank conversation with warehousing IT groups, 3PLs and shippers, you might feel as if we’re nowhere close to removing the human factor from the warehousing equation. Most sites today have manual planning processes supplemented by warehouse management systems, and very little automated decision-making is done. Most robotic systems that are in-place are treated as discrete function providers to supplement human labor, with little consideration for how robotic work can operate synchronously with human labor.

So instead of dreaming of a pie-in-the-sky state where all logistics operations are autonomous, let’s analyze some of the existing technologies for distribution centers and how they can drive tangible value in the short-term future.

There are a lot of technologies that you hear about every day that can help to streamline logistics operations: warehouse management systems (WMS), yard management systems, data replicas, digital twins, augmented reality, artificial intelligence (AI), robotic automation, and blockchain are some of the most recent popular technologies discussed by innovation groups. Likely, your organization may have already invested in some or all of these to make operations more efficient and deliver better customer service. However, how each of these technologies brings value is sometimes difficult to understand, as each is complex in its own way.

For the rest of this article, we’ll take some time to create technological stepping stones and disseminate how each technology operates, along with the value it brings. By the end, you should have a good understanding of which technologies can fit together to create a site with more capacity and less challenges.

Step 1: Setting a Foundation—Data Acquisition

Nearly all technology from the past decade relies upon accurate, reliable data. To ensure a site is able to harness all of its potential, it’s critical to first establish a ground truth for data. For your distribution center inventory, you need to know what it is, where it is and when it is needed. To do this, the foundational technology in place needed is warehouse management technology. This enables you to see:

• Future inbound and outbound orders (next 24-48 hours).

• Inventory on hand.

• Rough expected activities in the “work queue.”

A WMS is critical for all warehouse operations (truckload, less-than-truckload, e-commerce), and provides the foundation for nearly every other innovative system to grow.

Step 2: Normalizing Information—Data Replicas

Once you have your data acquisition system (WMS) in place, the next step is to make that data visible and accessible for additional development. Unfortunately, most WMSs are challenging to operate and not massively configurable once initially set up. In order to get around this, many customers are moving to a new breed of technology known as a data replica, which creates an easy-to-access database that pulls in real-time WMS data for querying, visualization and alerting. These data replicas offer a way to have complete operational visibility of everything inside of a facility.

As an interesting note on data replica technology, a lot of large organizations are starting to invest in building their own, or in creating a “thin client” that sits on top of existing systems to make data more available. As fun as that sounds initially, taking this approach deviates from the specialization of most organizations, bringing up all sorts of challenges around lifecycle planning, maintenance, feature support, and more.

Step 3: What-If Analysis—Digital Twins

Now you’re collecting data and can view it any way you choose. Next, it is critical to use that data to understand the future state of your distribution center. Many folks refer to this as “What-If” planning, and leverage digital twin technology to perform this task.

The reality is, the term digital twin is so overused that it’s tough to pinpoint exactly what it means. In warehousing, we refer to a digital twin as a mathematical model of a warehouse that analyzes all future-facing activities to predict what is likely to happen in the future. A good digital twin will account for labor, shipments, inventory availability, tasking, and space/resources. Some providers in this space focus on total warehouse activity orchestration.

To make this a bit more practical, digital twin technology can be used to effectively understand and manage shorts processes. By playing forward inbound and outbound inventory over time, a good digital twin will be able to call out that an order to a retailer going out in 18 hours time will be missing 72 cases of SKU 12345. This helps to provide decision support to planners, or potentially more robust algorithmic planners.

Step 4: Algorithmic Optimization—Artificial Intelligence Advanced Mathematics

Before we go too far, it’s important we make a key call out. Artificial intelligence (AI) means very little to anybody any more. Every marketing department tells you that you need it, but the reality is that AI by itself is not that useful for logistics operations. There are some AI technologies that can be useful for operational sub-tasks, such as forecasting or robotics control, but true AI is not valuable in many places. Instead, the term is used as a catch-all for advanced mathematics, which does a disservice to people who actually want to learn.

So, what is valuable with advanced mathematics is understanding the best way to actually execute work inside of a facility. Regardless of systems in place (machinery or people), all activities that occur in a DC can be quantified. That quantification can be paired with a flavor of mathematics known as constraint-based optimization, or convex optimization, to optimize different activity systems. When paired with a digital twin, this optimization technology can prescribe sequences of events, create a feasible operational schedule, and simultaneously minimize touches and labor while maximizing service levels.

In practice, advanced mathematics can be used to understand that there is an inbound scheduled at 7 AM that can be cross docked onto a drop load trailer for later in the day. This ends up saving a putaway and a retrieval while minimizing labor throughout the day across all different aspects of the warehouse. This example may be one of 100 activities occurring at any moment, and all of them can be optimized using advanced mathematics with the appropriate trade-offs.

Step 5: Labor Automation—Robotic Systems

The final stage in our technological roadmap is to automate many of the human processes that occur today. Whether it be using AS/RS, layer pick, or goods to picker systems, robotics can improve efficiency significantly. The challenge with robotics is that the capital expenditure is massive, so rollouts are challenging and often take a long period of time.

Many teams invest in robotic integration prior to the aforementioned steps in this article, but this often results in automated activities being just as unorchestrated as the manual activities. To reduce the likelihood of this, it is recommended to roll automated systems and robotics in as a part of a broader strategy around increasing capacity and driving efficiency.

Defining Your Roadmap

In conclusion, there are a lot of great technologies out there to improve operations in a warehouse environment. More so, they can be rolled out sequentially to drive efficiency across nearly any type of site. If you don’t have a WMS, start there. It can take many months to implement, and in some cases over a year, but it is worth it.

Once your foundational data acquisition platform is in place, building a data replica on top and then digital twin/advanced mathematics afterwards is a very fast process. All three in tandem can be implemented in less than six cumulative months. From there, the next steps taken with automation will depend on the business needs, budget and vision.

Logistics is no longer a cost center, but it is imperative that each technology along this journey show a marked return on investment. Before spending a dime, it is critical to understand both your business demand (now and future) and supply chain strategy to ensure complete organizational alignment.

Keith Moore is the chief product officer at AutoScheduler.AI (www.autoscheduler.ai), a provider of solutions that accelerate WMS capabilities with intelligent warehouse orchestration.

About the Author

Keith Moore

Keith Moore is the CEO of AutoScheduler.AIa WMS accelerator created to help orchestrate poorly coordinated facilities. He was previously chief product officer.

Latest from Technology & Automation