In the past five years, autonomous mobile robots, or AMRs, have gone from a niche, seemingly untested technology to a staple for many warehouses. While such robots have proved they’re here to stay, the industry still faces fleet management and other challenges in the coming years.
“Five years ago, AMRs were seen as a potentially breakthrough automation technology, albeit with questions about extended return on investment [ROI] periods and integration complexity, driven largely by limited real-world installations of the product class at the time,” said Vaithy Kandasamy, the global AMR portfolio manager at ABB Robotics.
At the same time, Interact Analysis significantly lowered its mobile robot forecast, citing a complex mix of geopolitical, economic, and industry-specific challenges, along with changes to its methodology in calculating market sizes. The market research firm reduced its 2025 forecast by $800 million, predicting lower growth in each of the world’s major regions.
As mobile robots begin to scale operations with large customers, it’s more crucial than ever for these providers to have a robust fleet management platform, built in-house or provided by a third party. These systems now do much more than just manage AMR movements; they also bring together data from various sources and interpret that data for users.
“We have an understanding of three data domains,” Florian Pestoni, the co-founder and CEO of InOrbit, told Automated Warehouse. “One is the business data, what work needs to be done. The other is spatial data. So, we understand the space where the warehouse is, or the manufacturing plant, or whatever it might be. And then we have very detailed robot data. Those three domains are usually completely disjointed, and connecting them is hard.”
Save the date for the 2026 Robotics SummitAMR fleets mature, as has customer interest
According to ABB’s Kandasamy, autonomous mobility has matured, and customers have taken notice.
“In recent years, technological advancements in vision systems, sensors, and batteries have enabled mobile robotics solutions to become increasingly flexible, collaborative, and efficient,” he said. “The AMR safety standards have also evolved and matured in the past five years, which increases the confidence in safe operation.”
In addition, many industries are facing the same labor issues.
“A lot of the end users we talk to are, regardless of what industry they’re in, facing similar challenges,” Pestoni said. “A lot of them have labor shortages, especially in the kind of tasks that are maybe more repetitive and less exciting. It’s harder to attract people to those kinds of jobs.”
Ati Motors, which is based in Bengaluru, India, has seen these issues pop up around the globe.
“We see a lot more inbound compared to before. I think awareness is high, and there’s way more intent today,” said Saurabh Chandra, founder and CEO of Ati Motors.
“This labor shortage problem isn’t going away,” Pestoni said. “What we see in users, where before, they might have only used robots in essential tasks that were super-repetitive and extremely well-defined. These are the classic applications of industrial robots, like in automotive, for example. Now, they’re thinking they need that [automation] for many more workflows.”
“Before companies look at KPIs [key performance indicators], they’re asking, ‘How do I keep my operations going?’ he continued. “They’re using robots so the people they do have can be used more efficiently.”
“They can kind of give them superpowers through these robots,” asserted Pestoni. “I think what has changed, especially recently, there’s more appetite to adopt and to adopt faster. It’s faster and in more places, so I think it’s a really exciting time to be in robotics.”
Fleet management providers work with many vendors
While the AMR industry has worked toward building robust interoperability standards, it’s a long and slow process.
“There’s an ISO standard that’s helping specifically with mobile robots. But that process is really, really slow,” Pestoni said. “Just getting to the standard being published is a multi-year process that involves lots of meetings around the world. Then, just because a standard is published, it doesn’t mean everyone has adopted it.”
Many end users want to use robots from multiple vendors to meet their specific needs, so this means companies like InOrbit need to act as a translator between all of these different systems.
“We do this all the time for our customers,” noted Pestoni. “We can put different robots on the same map and make them work together.”
Interoperability is more of a technical term for engineers, he said. It concerns the kind of protocols that robots from different vendors use to communicate. InOrbit approaches things more from the customer’s perspective, which Pestoni called orchestration. For them, all that matters is that the robots do work together.
“Taking that end-user perspective, what they really want is orchestration. Interoperability is one mechanism for that.” Pestoni said.
“We’ve connected many dozens of types of robots, and a lot of them have different proprietary protocols or APIs, or they use different versions of ROS [the open-source Robot Operating System], or they might support an emerging standard,” he explained. “Instead of trying to make everyone speak the same language, we’re like Google Translate for robots.”
AI is coming to help manage AMR fleets
InOrbit has been quick to implement the latest advances in artificial intelligence, like generative AI, into its fleet orchestration technology.
“I would say it’s everywhere in our product at this point,” Pestoni said. “And I don’t think we’re the only ones doing that.”
His company recently launched RobOps Copilot, which enables natural-language searches through a company’s data.
Typically, if something goes wrong in a warehouse, people would have to work like detectives to figure out where in the process the problem occurred. This means working with other stakeholders to track down information and sort through pages of data to find where the problem occurred.
“We basically put that power into the hands of the person that is acting as the detective,” asserted Pestoni. “We shorten this process that could take days or weeks to seconds.”
“I see it as an empowerment tool,” he added. “You don’t need to be an expert to get the responses that you need.”
AMR developers are also looking into ways to incorporate generative AI into their fleets.
“Large language models [LLMs] and generative AI are revolutionizing human-robot interaction with no need for specialized programs,” said Kandasamy. “Robot systems can interact with humans using simple spoken instructions or even by recognizing a hand gesture.”
“The system can learn continuously by itself to optimize the fleet performance and productivity,” he said. “Unpredictable or complex environments can be better solved with enhanced decision-making by AI, especially LLMs.”
“It can also enable smarter decisions such as fleet/mission management, fault diagnostics, predictive maintenance, etc.,” Kandasamy said. “This democratizes access to automation, removing the need for advanced technical skills and accelerating adoption across new sectors.”
Hurdles remain in AMR fleet adoption
The AMR industry has solved many of the problems that plagued it in the technology’s early days as a successor to automated guided vehicles (AGVs), which followed magnetic strips or QR codes.
“On the core technology side, we are pretty much sorted,” Ati Motors’ Chandra told Automated Warehouse. “Change management and adoption came in the large manufacturing giants, and we’re doing that last inch of automation. That becomes tricky, because it becomes custom. This is difficult to productize.”
AMR developers are also always working to make fleet deployments smoother and faster.
“Deployment time and integration complexity are two of the most significant hurdles in the adoption of AMRs, along with higher costs that come from such things, as the effort it takes to transition from other types of automation, adjustments required to the plant infrastructure, and securing the skilled labor needed to properly deploy AMRs,” Kandasamy said.
Finally, developers are working to create systems that are strong enough to scale, and end users may underestimate how complicated these systems are. Pestoni said many customers buy a few robots, use them for basic tasks, find them to be helpful, and then start to add more tasks on their own.
“They’re writing their own code, but their goal just is to run a factory or run a warehouse,” he observed. “So it might be more of a side quest versus it being the core of their job. So you end up sometimes with these systems that are cobbled together.”
“I love scrappiness, but when you start to scale, something that’s cobbled together will unravel,” warned Pestoni. “And, by then, the problem is, you’re in production. You’re breaking production. You’re not just trying an experiment.”
