Solving the challenges of robotic picking

The e-commerce market has been the growth engine of the retail environment, experiencing a 15% increase for grocery sales in 2015. One of the major challenges facing retailers is the ability to keep up with the fast growing demand coming from the online channels. This creates additional pressure on retail warehouses to find more qualified staff, whether for picking and delivering orders or for maintaining and expanding the existing warehousing infrastructure in a scalable way.

However, a recent GOV.UK survey showed that a growing number of jobs are being left unfilled because companies can’t find the right people with the appropriate skills. In addition, UK productivity on the whole has flatlined since 2008, a concerning trend given how other countries have dramatically improved theirs.

One solution to the challenges presented above is the adoption of robots working alongside human workers inside the warehouse to boost productivity, satisfy customer demand for fast delivery times, and reduce picking times. In the UK, Ocado has been a pioneer in the use of robotics and automation alongside its human workforce.

We are a net employer of 11,000 staff, but also create advanced hardware systems that help us maintain a very efficient operation. This in turn enables us to provide additional jobs and other employment opportunities, none of which would be possible without the technology we’ve developed.

SecondHands and SoMa

Ocado Technology has been one of the first companies to realize the positive impact of advanced robotics in the e-commerce market. In addition to creating a hive system for our Ocado Smart Platform, we’ve also partnered with several universities on two robotics-related projects funded by the European Union: SecondHands and SoMa.

The SecondHands project is interesting not only from the perspective of what we will learn along the way, but also because we plan to build many highly automated CFCs worldwide (for Ocado and its OSP customers). Therefore, the SecondHands robots would help us maintain the CFCs, and ultimately help automate their construction too.

While SecondHands focuses on designing a robot assistant for industrial maintenance tasks, SoMa is exploring the utility of strategies that embody a soft manipulation approach. Typically this entails using a robotic hand that exhibits controllable compliant (i.e. spring-like) behaviour.

Given that Ocado offers over 48,000 different items on our store that vary in size, shape, weight and span from rigid to highly deformable, the SoMa robotics teams must solve many unique challenges in addition to the classical robotic manipulator problems.

Recently, the team has been working on a benchmark framework for the evaluation of soft manipulation systems used for commercial purposes. The framework takes into account several key attributes and assumptions, including the asymmetric and deformable nature of the item to be picked and the damage-free handling of the product or packaging.

Performance-wise, a successful grasp in the Ocado use case refers to robustness (i.e. a secure and reliable grasp), a low level of damage or bruising observed, and a short pick-and-place cycle time.

The benchmark framework was recently presented at a workshop (part of the International Conference on Intelligent Robots and Systems held in South Korea) to a large audience of robotics experts from around the world.

The presentation was titled “Systematic evaluation of compliant under-actuated soft manipulators in an industrial context — the Ocado use case” and included a description of the software and hardware setup being used to test several hypotheses concerning the efficacy of various pre- and post-grasping strategies employed with robot hands that are consistent with the soft manipulation paradigm.

If you’d like to know more about SoMa, visit the project’s website and follow the team on Twitter.