First came the realization that a robotic system’s software—more so than its hardware—make these systems have an impact on materials handling.
Software is what is constantly calculating the optimal routes for a fleet of autonomous mobile robots (AMRs) to take, and it’s what gives robotic piece-picking arms the smarts to know how to grasp and place items.
Behind every good robot system, there is some robot operating system or fleet software not only doing basic control or navigation, but in some cases applying artificial intelligence (AI) and machine learning algorithms that allow these systems to get better by doing.
Now we have another sign of software’s importance to robotics: platforms that coordinate tasks between multiple AMR fleets and traditional automation, while simplifying integration to warehouse management systems (WMS).
Analyst firm Gartner calls this emerging arena “multi-agent orchestration software platforms,” with the “agent” part of the moniker noting that it also involves integration with traditional automation like conveyor controls, says Dwight Klappich, research vice president with Gartner.
Klappich says recent studies, including one by Gartner on supply chain technology needs, and research into plans for warehouse robotics he collaborated on with Peerless Research Group (PRG), made it clear that healthy uptake for AMRs and other forms of robotics was on the way.
The PRG research found that while 86% of respondents will look to make further gains with AMRs by expanding the use of one AMR fleet, the study also found that 92% plan to expand robotic use cases, a finding that points to increasing use of multiple types of AMRs. Heterogeneous environments, Klappich concludes, are going to be more likely going forward.
“The minute companies start to have heterogeneous fleets of robots, we start to have a challenge,” says Klappich. “If it’s just one type of AMR, and you are running a WMS, then most companies are just going to find some way to integrate that AMR fleet with the WMS directly, but when a company starts having two, three or five different types of robots, there starts to be problems in terms of integrating to all of them efficiently—and orchestrating which systems get what work tasks and when.”
As a result, Gartner estimates that by 2026, more than 50% of companies deploying intralogistics robots will have a multi-agent orchestration platform.
As an emerging niche, vendors in the multi-agent orchestration platform arena vary in their capabilities, though share some commonalities. These platforms are more than just a set of pretested application programming interfaces (APIs). They generally seek to shield end-user organizations from having to hard-code the links between systems.
According to Klappich, some vendors in the orchestration platform space focus more heavily on the integration challenge, while others stress orchestration features. Both capabilities, Klappich adds, are part of this emerging space.
“First, there’s need to speed up the integration piece of it, so it doesn’t take months to onboard a new system if you can get the robots in two days,” Klappich says. “And, there’s also this need for more business logic around how to best assign work, how to parse out the work and the tasks to different systems, to create this orchestrated flow of work. The real value lies when you start to get to that second level—the orchestration piece.”
Other analysts also see a growing need for platforms that coordinate AMRs. Interact Analysis, for example, estimates the segment it calls multi-fleet orchestration platforms will grow at a CAGR of 138% through 2027.
Among the WMS and warehouse execution system (WES) vendors addressing robotics integration is Softeon, which recently acquired warehouse technology and implementation firm GetUsROI and its software-development arm.
That brought with it an orchestration platform, which works on a publish-and-subscribe paradigm for sharing information that different systems need to interoperate, including AMRs, but also other forms of automation or human-centered processes like RF-based picking.
Dan Gilmore, CMO for Softeon, says that even though today only a modest percentage of companies have more than one AMR system, many more have traditional automation, and nearly all have WMS processes and methods such as RF-based picking, so there’s a need to evaluate easier ways of making all these resources work together, even if your DCs aren’t multi-fleet environments yet.
“We’re amazed at the many organizations, including relatively smaller enterprises and many third-party logistics providers [3PLs], who are now looking at mobile robots because they just can’t find and keep enough people to run their warehouses without more automation,” says Gilmore. “As more companies start off on their robot journey, it’s important that they consider how they will support the needed integration and collaboration between their systems over the long run, and in a way that gives them the flexibility to rapidly adapt to new equipment and new processes.”
Mark Fralick, Softeon’s chief technology officer, describes Softeon’s orchestration platform as a “composable” integration approach, meaning non-programmers can design workflows with it. The platform, he adds, works on a publish-and-subscribe message broker architecture that lets various systems or “actors” subscribe to needed information.
The platform also contains a utility that translates the messaging formats among diverse systems, so users don’t have hard-coding of the links to a WMS or a warehouse execution system (WES). Instead, the focus can stay on how to design the best material flows and configure the handoffs between systems.
“We think about everything in terms of actors, actions and objects,” Fralick says. “And, under a concept we call activity streams, you got these actions coming in, and these actors who are part of the platform, and from there you can distribute to whichever actor has the capacity and capability to carry out a specific action. It might be a robot or AMR. It might be another form of automation like a cube storage system. To us, it’s just another actor that’s helping the overall flow of the facility.”
Perhaps the biggest benefit of multi-agent orchestration software is the speed of onboarding new systems, says A.K. Schultz, co-founder and CEO of SVT Robotics, a key vendor in the multi-agent orchestration space. With no projects to write custom interfaces, it’s faster and easier to gain the benefits of new systems.
“I think speed to value is the thing that everyone really, really wants,” says Schultz.
Schultz explains that SVT’s SOFTBOT platform abstracts the technical details of how different WMS solutions define concepts like a “pick” or a “replenishment” so that all a robotic or automation system needs to do is to talk to SVT’s platform, rather than understand the way specific WMS solutions define the world.
Schultz says one WMS vendor’s data definition handbook is about 1,400 pages. The better way is a common language, he believes.
“We tend to sit at the center of the data exchange and orchestration at a business logic and business outcome level,” he says. “For instance, when an order comes into a WMS, we convert that order through a connector into a common language, if you will, and extract the specific information that is needed by a particular AMR system, in a way that it likes to consume data. Then the AMR system controls its fleet.”
SVT’s platform also uses an information bus mechanism for communication and features Vizio-like tools for designing and adjusting workflows. The platform also has “parsing bots” that can govern task assignment decisions with simple “yes/no” rules, like whether a system is capable of handling a certain size SKU or load.
The workflow automation part of SVT’s platform is crucial, adds Schultz, for easily establishing workflows that result in a smooth overall material flow and permit rapid adaption over time. “No. 1, you’ve got to get everyone speaking the same language, but No. 2, you then have to do something with that information to create the desired workflows and outcomes,” Schultz says.
Companies using SVT’s platform include Kenco Group, a major 3PL. Overall, says Schultz, 3PLs have strong interest in robotics orchestration and integration software because they’re rapidly adopting AMRs, since they don’t take much infrastructure or upfront capital to deploy.
On top of that, 3PLs are often running multiple WMS solutions, so traditional integration is costly to build and maintain. “What we’re trying to do is de-friction all of the work and the challenges involved in integration, which is really hard to do and requires extensive, specialized knowledge, without a good platform,” says Schultz.
GreyOrange has roots as an AMR vendor, but its GreyMatter software has evolved to handle multi-agent orchestration. Samay Kohli, co-founder of GreyOrange, says a multi-agent orchestration platform should simplify integration, but also be strong on order processing capabilities and have a holistic view of inventory and inventory consumption across different areas to optimally orchestrate resources to meet service level agreements (SLAs).
“The system has to be able to understand and orchestrate inventory and orders in a manner that ensures the right outcome,” Kohli says. “It has to be able to provide work to all the people involved in the fulfillment process to ensure the optimal output.”
For example, a large format bot could take pallets from receiving to a buffer or active picking area where there might be other automation for replenishment or order picking, and perhaps end-of-line parcel sortation. The orchestration platform would look to optimize the whole flow to hit SLAs.
“In this entire operation, a multi-agent orchestrator should be able to look at all agents within its disposal, tasks from inbound to replenishment to order picking, the number of orders and their SLAs at hand and the number of people involved in different processes to understand capacity, flows and efficiency to ensure that SLAs across all these processes are met, while all resources are utilized optimally with little to no waste,” Kohli says.
Companies using GreyMatter include apparel retailer H&M. Overall, adds Kohli, even if a DC isn’t yet multi-agent, an orchestration platform is a way to “derisk” their automation future. “Using a multi-agent orchestration platform gives companies the freedom to choose the ‘best-of-breed’ fulfillment technology that fits their needs without restricting them to a specific vendor,” he says.
CoEvolution, another vendor in this space, offers an integration layer and also emphasizes its fleet management capabilities. The CoEvolution platform handles higher-level fleet management functions, including path planning, traffic management, task assignments and scheduling for smart mobile assets, says Lijun Zhu, CoEvolution’s CEO and co-founder.
Zhu explains this central fleet management is key to tightly orchestrating the moves and tasks of different robots, so one system isn’t waiting on another. The system knows all mobile robot characteristics like payloads, speeds, dimensions and turning radius constraints.
“The bigger value we provide is this overall orchestration and optimization, done holistically in one platform, not separately,” Zhu adds. “If there are direct handoffs of materials between robots in a shared workflow, we’ll make them connect seamlessly, so there’s less idle time.”
For example, says Zhu, using the platform’s orchestration, an autonomous lift truck can pick and retrieve pallets from high-density, multi-level racking, and set it in place for horizontal transport to a downstream location by a lower-cost AGV or AMR.
The system can also alert fixed automation, like a depalletizing cell, that a pallet has arrived and is ready to process. “Basically, we let different robots perform the tasks they are best suited for, and then let them collaborate with each other more efficiently and cohesively,” says Zhu.
Vendors with roots in WMS also may be tackling multi-agent orchestration. They see it as a customer-driven move since DCs are becoming more automated, not just with AMRs, but other smart automation like packaging machinery.
For example, Synergy, which offers the SnapFulfil WMS, recently introduced SnapControl, a multi-agent orchestration platform that coordinates workflows to be carried out by devices or robots, automatically allocates tasks and workflows, and evaluates the robotic devices or other systems that best match specific operations, says Smitha Raphael, chief product and delivery officer for SnapFulfil.
The platform has two key layers, explains Raphael: configuration and a rules engine. The configuration layer is where the platform defines common tasks like picks and maps that to how AMR systems view missions.
The rules engine, adds Raphael, makes SnapControl responsive to feedback from robotic or automation systems, like if a system goes down, to decide the next best resource to tap. “The rules engine is just directing work; it’s responding to change. If a system is unavailable, the rules engine basically says, ‘OK, then I can direct work to this other system that can carry out that same task.’ It’s like a conversation,” she adds.
One SnapFulfil user has already deployed SnapControl to manage allocation of work to AMRs from 6 River Systems, and routing of work to carton right-sizing equipment from PackSize, says Raphael, while others are evaluating ways to use SnapControl.
Last fall, Körber Business Area Supply Chain, which offers WMS and other supply chain software, announced the launch of its Unified Control System (UCS), a solution for configuring and managing warehouse technologies, including AMRs and robotics, as well as voice picking and WMS workflows. An early user of the solution is Boxy, an e-commerce 3PL in Hungary, which in addition to Körber’s WMS, is using UCS to coordinate the tasks for AMRs from two different vendors, Geek+ and Libiao, as well as pick- and put-to-light solutions.
“We see the future as having many vendor solutions in a single facility, with newer solution types having to co-exist with conventional automation that may already in the facility, and then also with human-centered processes and systems, like voice,” says Joe Couto, executive vice president of robotics and 3PL at Körber. “As a result, you really need a unified solution that will allow you to manage all of these systems.”
Couto contends that Körber is well positioned for multi-agent orchestration because of its domain expertise across WMS, voice, warehouse control system (WCS) software, as well as helping its users adapt their warehouse processes to operational challenges like demand peaks. With the addition of UCS, he explains, companies have a platform to coordinate all their key resources—robotics, traditional automation, and human labor—in a way that’s adaptive to ongoing changes like ramping up capacity for peak seasons.
“We also see voice as continuing to be an important platform, from the standpoint of if there is a peak in activity, sometimes bots can be applied to handle those peaks, but in other cases, it may be more effective to handle peaks with human labor and voice,” Cuoto says. “So, UCS can be a common platform that enables scaling with voice or scaling with bots.”