Despite economic headwinds, robotics will continue to spread across warehouses, distribution centers, and other facilities this year, according to RightHand Robotics Inc. The Charlestown, Mass.-based company provides software and services for robotic piece picking and fleet management.
The market for warehouse automation will experience a 19% compound annual growth rate (CAGR) between 2024 and 2027, according to Interact Analysis.
The underlying challenges of labor cost, availability, and turnover are continuing to drive demand for automation in warehouses and beyond, said Brendon Bielat, vice president of product and marketing at RightHand Robotics. He has experience from Google and Amazon Robotics.
“We’ve talked with a lot of customers that were very interested in automating and wanted to experiment with one or two robots to see if they lived up to the claims,” Bielat told Mobile Robot Guide. “A lot of those same customers are coming back and saying, ‘OK, now how do we scale this?’ I think that conversation is indicative of the entire industry.”
“The second piece is that customers that are buying robotics aren’t typically robotics experts — obviously, their primary business is that they’re trying to fulfill orders for retail or e-commerce,” he added. “They don’t want to spend the time maintaining, configuring, or programming robots. That’s an intimidation factor that can stand in the way of adoption.”
AI popularity opens the door for robotics
While generative artificial intelligence isn’t yet a part of most production, interest in AI has changed how potential customers see robotics, Bielat said.
“The idea of having robots with AI in day-to-day warehouse operations is more approachable because AI is now a part of our everyday language,” he said. “Generative AI has seen some interesting use cases within robotic piece picking, with companies saying a user could be able to tell a robot to pick an orange.”
“It could be useful for things that were largely done manually and might take many hours of work for an individual or team to program,” noted Bielat. “Things like automated labeling of images that are used in machine learning models. In that case, I think you’re going to see some acceleration of the ability to train models.”
In addition to robot training and simulations, progress in machine vision and cheaper sensors have lowered price points and helped customers with returns on investment (ROI), he said. “You now have a much more compelling business case,” said Bielat.
Robot grasping needs more than suction cups, says RightHand
Several robot vendors have relied on suction cups rather than more precise poses and fingers, but vision and software are necessary for handling a wide range of products, explained RightHand Robotics.
“For small SKU sets, it’s easy enough for a customer to find the ideal end effector,” Bielat said. “That makes sense for larger items. But there’s a pretty big difference between picking something like a small roll of Chapstick versus a ream of paper. We’ve come a long way in figuring out how to combine vectors and fingers.”
“There’s always the option for companies to swap out the grippers, but they may not have the luxury of time or be able to integrate all the data for that,” he added. “We’re coming up with some pretty clever ways to handle very large ranges of SKUs.”
Find the best mix of robotic and manual picking
Regardless of the hardware or software used, reliability is a key metric for picking, acknowledged Bielat.
“A reliable range is not that a robot can pick something five out of 10 times,” he said. “We’ve had some customers come to us with unique applications. In some cases, it’s not a good use of a robot, and in others, we can do a lot more. There’s an upfront process where we sit down with customers, understand their workflows, and help them optimize for robots. You’re still going to have some manual pick stations for certain items, workflows, and exceptions.”
Bielat cited the example of a RightHand general merchandise customer with tens of thousands of SKUs. The company studied the workflows, which turned out to need an industrial robot rather than a piece-picking system.
“It’s difficult to have a simulation that can accurately depict all of the various factors,” he said. “We’ve run it through all the models and found where we can have a 90% or more success rate, and continually improve that percentage over time. There are others where we’d say, ‘Hey, it would be better for you to send this to a manual station.’ On Day 1, our system can pick some great percentage, and we have continuous software updates throughout the year.”
Robotics to pick up soon and fast, says RightHand’s Bielat
Robotic picking is starting to be paired with loading autonomous mobile robots (AMRs) and at the edges of automated storage and retrieval systems (ASRS), said Bielat.
“We’ve been doing a ton of work, both on the software side and on the hardware side,” he hinted. “As the industry sees things like ASRS and AMRs be commoditized, it will look at where else to get the speed or flexibility. More customers are ready to take on an automation strategy.”
RightHand Robotics is starting to observe more robot-to-robot picking, as well as pairing piece-picking systems with quality control, labeling, or weighing upstream or downstream, Bielat said.
“The higher your install base, the faster you’re going to learn,” he added. “The technology is only getting better, as is the integration. Once the capital markets pick up, we’ll be on the edge of a very steep increase in adoption across the industry.”
“A lot of customers want to try one or two robots — they don’t go from zero to 100, they go from a few to 20 to 100,” said Bielat. “The sooner that you can get on track of seeing how they work in your facility, the better you’ll be able to quickly implement the technology when that explosion of growth happens.”
