The objective sounds relatively simple.
On a designated day in a single DC, pick, pack and ship a particular set of SKUs in a range of desired quantities to 24 retail stores. Or 50. Or 100.
That’s what a DC does. Right? Yes and no. But there’s a problem: All those new robots.
Fulfilling orders is not all that simple if several types of robots are involved. That list could well include autonomous mobile robots (AMR), piece-picking, goods-to-person, robotic put walls and even shuttles in an automated storage and retrieval system (AS/RS).
While all operate in the same facility, they don’t typically interface with or talk to each other. They are islands of automation. That’s a blast from the past that wasn’t a compliment in the 1980s and 1990s, either.
Furthermore, there’s the matter of pacing and sharing the workload across those robots so store shipments can be made on schedule. Quite simply, how do they get the work done?
Good old fashioned inventory management is both part of the challenge and central to the solution to integration of those robots.
“You want there to be enough SKUs distributed across various buffers. And you don’t want to have too many items of each SKU or too few, limiting flexibility,” says Grant Beringer, vice president of integrated systems at Swisslog.
None of this is a hypothetical or a one off. Beringer and others have been asked more than once lately to integrate different types of robots.
As it turns out, this can be a bit of a real-world conundrum. Welcome to the new world of heterogeneous robotics integration. (Not that we need yet another acronym, but we’re going to call this HRI just because it rolls off the tongue much more easily than that tongue twister.)
HRI really is a conundrum. If you’re looking for the root of HRI solutions right now, start with Dwight Klappich, research vice president and Gartner fellow, and Kevin Kuntz, head of supply chain for Gap.
Klappich is the one who saw a lot of robots running around completely independent of each other, and pinned a name on what has to happen next to make them work together.
Kuntz is the guy who is in the thick of what’s next. He’s learning how to integrate different robot types at a Gap DC in Gallatin, Tenn. (Modern Materials Handling, March 2023.)
Last year, Gartner did a survey about user wants and needs for supply chain technologies. About 96% said they currently use or plan to use robots. Similarly, 93% plan to increase the size of their robotic fleets. And 94% plan to pursue other robot use cases in their facilities.
“By 2028, 50% of large enterprises will have adopted some form of intralogistics smart robots in their warehouse and manufacturing operations,” says Klappich.
As you already know, not one type of robot fits all applications. In fact, Klappich goes on to say there are 34 different categories of intralogistics smart robots. They already touch operations from truck unloading and depalletizing to putaway, picking, sorting, packing and shipping. Not to overlook replenishment and inventory management.
All of this is going to require orchestration. Klappich talks about a multi-agent orchestration platform that assigns, prioritizes and coordinates what are now mostly islands of automation. That’s a tall order, as it turns out.
In his facility in Gallatin, Kuntz is living Klappich’s dream.
He already has automated storage shuttles (TGW), robotic put walls (Kindred), robotic truck unloading (Boston Dynamics) and a dedicated returns handling robotic storage and retrieval system (Exotec) in place.
The key to innovation success, Kuntz says, is identifying quickly whether a technology can be integrated into the flow. At this point, a warehouse execution system (Vargo) makes that possible in Gallatin.
Or as Beringer of Swisslog points out, the challenge is to integrate each robot successfully so it performs well not only on its own, but as an integrated member of an automated handling system that fills orders accurately and on time.
Fortunately, a range of suppliers is in hot pursuit of solutions.
For this story, we talked with a robotics supplier (RightHand Robotics), a platform supplier (SVT Robotics), a solutions provider with an HRI platform (Hy-Tek Intralogistics) and a supplier of hardware, software and solutions (Swisslog). There are many other suppliers of various partial solution sets, just so you don’t think these are the only ones.
As you already suspect, each has its own perspective on HRI, for the most part built on the supplier’s current niche. And no one can go it alone. Both conditions make it clear just how new HRI is.
“The core story of heterogeneous robotic integration is all about making physical and digital connections,” explains David Schwebel, head of sales and strategic business development at RightHand Robotics.
On the physical side, each robot needs to do its job as a stand-alone piece of equipment. In addition, many of these robots need to be able to physically interface with other materials handling equipment such as conveyors or automatic guided vehicles. In other instances, robots need to physically interface with other robots. This is where it can start to get hairy.
For instance, a piece-picking robot needs to select and remove eaches from a tote on an automated storage shuttle. But as Beringer explains, the proliferation of robots in DCs means the number of permutations of robot combinations can accumulate in relatively short order.
Each robot has its own software and APIs, explains Zac Boehm, vice president of innovation and technology at Hy-Tek Intralogistics. None were written to interface with other robots or additional types of automation.
But to be effective, new, digital connections have to kick in. The different robots have to communicate with each other to move and manage the needed inventory. This digital connection is where stand-alone robots literally take on a new life with heterogeneous robotic integration.
As Boehm explains, there’s a problem to making that happen. “To communicate horizontally across equipment, you have to first communicate vertically to any of three systems—warehouse control systems (WCS), warehouse execution systems (WES) or warehouse management systems (WMS).”
Sounds simple enough. But it’s not. Klappich may have a name for it—multi-agent orchestration platforms—but they are just now being developed. As it turns out, it’s not all that easy to assign, prioritize and coordinate tasks across islands of automation.
The core issue is fairly straightforward, explains Schultz of SVT Robotics. It’s the difference between people and robots. Which is considerable.
People can fill in for themselves the tasks needed to hand off an item to a tote, for instance. However, robots need to follow written code to translate every desired task into an action, says Schultz. But writing that code is only one step, and is typically done by custom developers to connect each point-to-point solution across technologies and with enterprise software.
There’s also the matter of determining priorities for robots in each moment. That’s where the WMS, WCS and WES come in. Each in its own way manages equipment and inventory based on a set of priorities created to fill orders for the day. Only trouble is the communication layer between robots and those software systems.
That’s where these multi-agent orchestration platforms come in, says Sandy Stephens, executive vice president of software and consulting at Hy-Tek. He explains these platforms are the clearing house for data exchange between all agents (multi-agent) in the automated systems. That’s the integration everyone wants.
Schultz puts it this way: “An orchestration platform does not just send messages. It also makes decisions and sets priorities for each piece of automated equipment. It stitches together all the ecosystems in an automated system.”
And while the platforms from Hy-Tek and SVT Robotics are different, they are both trying to agnostically enable robots of all types to interact in real time to maximize workflows.
Orchestration platforms such as these also address a common sour spot in materials handling automation integration—how long it takes for software to be written to create the integration in the first place. That’s a common source of system launch delays.
But with pre-set platforms, much of that software is already in place and ready to make full integration a reality in much less time.
If HRI sounds intimidating, that’s because it is. And while we may be at the early stages of making it manageable, there are a lot of great minds working on the problem. Your robots will not be alone.