When Ocado Group launched in 2000, its founders had an ambitious goal to change the way the world shops for groceries. Its mission was — and still is — to solve the complexities of the online grocery channel and build a scalable, profitable model for ecommerce.
Hatfield, U.K.-based Ocado is certainly on the right track. The company‘s retail arm now has a geographic coverage of over 80% of households in Britain, supported by over 20 customer fulfillment centers and with 13 online grocery retail partners.
The success Ocado has seen over the past two decades has a lot to do with the automation it designed and developed in-house to meet specific grocery challenges. With a high number of SKUs and other strict rules around temperature management or the bundling of products in a bin, there’s a lot of complexity within the logistics of an operation.
To be able to win in the low-margin online grocery game, a retailer must fulfill orders at breakneck speed while adhering to quality-control and food-safety mandates. It’s a scenario that is so specific and so sensitive, it requires a unique technology approach.
Matt Whelan is the automated storage and retrieval system (ASRS) engineering director at the group’s Ocado Technology development arm. During the early discovery process for automation, Whelan was a software developer on the simulation team responsible for testing hardware systems prior to purchasing anything. The team wrote artificial intelligence algorithms to figure out how to move robots around fast enough to fulfill grocery orders.
“In the early days, we partnered with third parties for the hardware development because we hadn’t previously developed our own hardware,” Whelan told Automated Warehouse. “We had reasonable success with that. But when we wanted to make newer, higher-reliability, lower-cost versions, we thought we would do a better job ourselves.”
Building a better ASRS
Ocado’s 500 series on-grid robot was the first bot completely developed in-house, from the firmware through to the mechanical design and everything that goes with it.
guided cobot arms. Source: Ocado Group
The next ASRS version, the 600 series, was also developed in-house with a similar look, but it is one-third the weight of its predecessor. The weight reduction, a result of 3D printed parts, adds value in the form of energy efficiency. Lighter grids are also easier to install and cost less to use in existing facilities.
“When we started working on the 600 series, we landed on this understanding that explicitly driving down the weight of the bot itself has this multiplicative effect on the value that it brought to the system,” Whelan said. “So that’s what’s taken us into all of the advanced manufacturing methods using the industrial 3D printing, the topology optimization, carbon fiber, as well as pulling weight out of the system, which is hugely important for what we do.”
Creating such innovation requires dedication to the development of the many layers of automation. An Ocado Technology engineering team that included about 350 people used special design and modeling techniques to create the lightweight bots.
In addition, Ocado engineers design the inner workings of the machine, including motor drive circuitry, firmware development, and motion control.
“So that is, how to make a certifiable safe piece of circuitry that can spin a motor around,” said Whelan. “And then how to convert that into a motion profile that a bot can accurately move along with various sensing strategies. And then up to how to communicate where [it is] to a central system that is orchestrating how everything moves around. So, it’s quite [a lot of] layers to the system.”
Organizing the hive with AI
The mobile bots sit on top of a grid moving around in swarms within 5 mm (0.1 in.) of one other at speeds of 4 m/sec. (8.9 mph) – forming “The Hive.” Together, the hive, the grid, the bins, the bots are a complete fulfillment offering called the Ocado Storage and Retrieval System.
According to Whelan, AI is central to the orchestration of the robot fleet, figuring out optimal routes for picking SKUs and placing them in totes.
“There’s all kinds of clever balancing optimization algorithms that drive that system,” he said.
A new capability of the ASRS is the on-grid Robotic Pick system. Vision system-driven robotic pick arms mounted on the grid collaborate with the bots. There’s a huge number of AI and machine learning systems within that picking process.
In addition, AI and vision systems help operate global fulfillment sites. CCTV cameras are deployed above the grid with AI systems behind them, enabling Ocado team members to monitor and manage the grid from anywhere in the world.
If a bot has a problem, for example, they can block an area of the grid off.
Ocado Group expands automation beyond groceries
In 2017, Ocado expanded operations outside of the U.K., with Bon Preu in Catalonia, Spain. In 2018, it announced an exclusive partnership with U.S. grocery retailer Kroger. The two companies are building customer fulfillment centers across the U.S., ranging from large facilities to micro-fulfillment.
The company established a new business unit in 2023 – Ocado Intelligent Automation (OIA) – to drive the commercialization strategy to expand to markets outside of grocery, including consumer packaged goods (CPG) and healthcare. Given the work that has gone into developing the Ocado Smart Platform and the experience the group has in building fast, reliable automated systems, it’s a solid expansion plan.
“We’ve been operating massively automated warehouses for 20-plus years, and we know what we’re doing,” Whelan said. “As we compete in new industries, we’ll see what challenges that throws at us. … But we will be pushing the boundaries in terms of performance and general cost of ownership.”