January 31, 2017      

One of the biggest hurdles to more widespread supply chain automation is the ability to manipulate a wide variety of objects. In Europe, robotics research into soft grippers is making progress as part of a four-year program.

Ocado Technology, the research division of online grocer Ocado Group Ltd., has been developing soft manipulator technology with European partners. The British company has also been working on its SecondHands humanoid robot, as well as other warehouse automation.

The SoMa project was funded through the EU’s Horizon 2020 program and includes collaboration between Ocado, the Technical University of Berlin (TUB), the University of Pisa in Italy, the Italian Institute of Technology, the German Aerospace Center (DLR), the Austrian Institute of Science and Technology, and Disney Research Zurich.

Business Takeaways:

  • Ocado, the U.K.’s largest online grocer, is actively developing different robots for its warehouses, including soft manipulators for common household items.
  • With partners across Europe, Ocado is conducting research for the SoMa consortium and doesn’t expect the Brexit to affect its Horizon 2020 funding.
  • The technical challenges to be solved and potential market are both large enough that Ocado’s head of research doesn’t SoMa to be duplicative of other robotics research.

Consortiums reach across the Channel

“A few years ago, we were busy forming our own consortium for SecondHand,” said Alex Harvey, head of research and simulation at Ocado Technology. “We got invited to this consortium, which we formed with the Technical University of Berlin, so we were happy to be the U.K. research contribution to SoMa.”

“Most partners have contributed research,” he explained to Robotics Business Review. “Some have contributed hands, others the robotic arm. We’ve conducted research at our facilities into generic parts planning and vision-guided manipulation.”

SecondHand successfully passed its Year 1 review last summer, and we’re working on an updated integration plan for the five-year project,” Harvey said. “An Ocado lab warehouse has been built — with our own money, not Horizon 2020’s — and it’s ready and waiting for robots from the Karlsruhe Institute of Technology, which should be delivered in the next few months.”

“SecondHand is vision-based manipulation for rigid items, while SoMa’s RBO Hand 2 is the complementary research for deformable items,” he said. “All the research projects are to enable us to pick the full range of products for our uses.”

Although Horizon 2020 funds are involved, the so-called Brexit has not affected the SoMa robotics research.

“Happily, the project officer has confirmed that we’re fine,” Harvey said. “The contracts were all signed 18 months ago, and we’ve been told that we can still file applications.”

“Officially, we’ve been told that nothing has changed, but the concerns are very valid,” he said. “Politicians could still do something. Human bias always plays a role. When evaluating European projects, U.K. consortium members might not review participation as favorably as possible.”

“Horizon 2020 projects are a really great funding stream [for robotics research],” Harvey noted.

Robotics research proceeds

“The challenge is not finished yet — we’ve made Mark 1 progress,” Harvey said. “We have a good level of integrations, and some very basic environmental constraints such as the side of a fruit tray.”

He enthusiastically said that work on the SoMa robotic arm and end effector “gets much more exciting from here. We’re devising simplistic science experiments. We’re already finding fascinating outcomes but can’t share them yet.”

Ocado has a very practical goal in mind for the consortium’s research. “We have 48,000 types of items in our warehouses,” said Harvey. “It’s a colossal range, from packs of Evian bottled water to kindling for barbeque.”

“The complexity and variability includes a whole subset of SKUs that are soft or could be damaged, like a tomato,” he said. “That’s a subtly different challenge from an egg, where the mechanical properties of the surface are different. We’re building up to that complexity.”

“SoMa is really interesting and is using environmental constraints, which humans use all the time,” Harvey said. “The fact that it’s focused on our use case makes it more interesting.”

“The twist on soft manipulation with constraints is both novel, and there’s scientific benefit of moving [forward] on the overall state of the art for manipulation.”

“We’re about 18 months into a four-year project,” Harvey said. “We’ll be trialing more grippers from Pisa and DLR and hope to give feedback to those institutions so that we can develop Version 2 soft grippers.”

Ocado also plans to test more objects with SoMa with its e-commerce order-fulfillment warehouses in mind. “We’re looking to explore more complex scenarios,” Harvey said.

Commercialization and competition

Although several other robotics research organizations are working on soft manipulation, Harvey wasn’t worried about competition.

“For Ocado, having a robot that could pick and place 48,000 items with a general platform — it’s a hard problem,” he said. “The problem space is big enough for lots of people to be looking at the complexities without much overlap.”

Robotics research prototype of a soft gripper.

Ocado hopes that its robotics research into a soft manipulator will benefit both itself and society.

Unlike some other teams, the fact that Ocado is both a research partner and an end user gives it an edge, according to Harvey.

“It’s important for us in shaping the item,” he said. “There have been instances where we’ve bought things that weren’t ideal for our purpose. We’ve had to struggle with customization to get to optimal performance.”

“Efficiency is essential in our business,” Harvey said. “Eking out every last cent of performance — that last few percent is where all the profit is.”

“Getting close is quite easy; getting something that really works is really difficult,” he added.

More on Robotics Research and Logistics:

Don’t fear the robot

“This technology is truly transformational, in terms of capital and operating expenses,” Harvey said. “It’s not putting people out of job, but aiding the operational aspect of jobs in warehouses that currently involve people. Imagine the time saved by not having to walk around just picking objects.”

“With packing or bagging groceries, there’s the challenge of placement, repacking manually,” he said. “That’s levels of complexity above SoMa, but for doing things on a day-to-day basis, we’ll have to incorporate elements from SoMa.”

“We’re trying to move forward with manipulation in a generalized manner of household objects. Sure, these capabilities benefit Ocado, but they have a wider societal benefit,” he observed. “If we can develop end effectors to manipulate consumer goods found in home — in the kitchen, the bathroom, wherever — we can develop manipulators that could eventually help the elderly or people with disabilities.”

Harvey disputed the popular belief that automation will lead to increased unemployment.

“Ocado is very passionate about its employees, who give it its culture and make us who we are today,” he said. “We also value our customers, and our value proposition will always balance the two.”

“We must provide the benefits of automation to our customers, but we’re a happily growing organization,” Harvey asserted. “Ocado overall has more than 11,000 employees, and Ocado Technology has about 1,000 people. We can offer roles to people where automated systems are deployed; we haven’t needed to make anybody redundant.”