Dr. Graham Deacon
Soft Manipulation in the Delivery of On-line Grocery Orders – The Ocado Use Case
Ocado is the world’s largest on-line only grocer. We already employ high levels of automation throughout the business and a natural next step is to employ robots to achieve the mass customization of the assembly of grocery shopping baskets.
Robot manipulators are typically deployed in circumstances where the environment is engineered to be predictable and there are a small number of parts that are a regular shape and/or have CAD data describing them. In our case we want to handle ~48K different objects whose locations are difficult to constrain, there is wide variation in size, shape and materials, no CAD data, there are not only deformable but also easily damaged items as well as those whose load distribution changes when they are picked up.
This talk will describe how we expect soft manipulation to help us address these challenges and how we assess the performance of inherently compliant end-effectors in our use case.
Graham Deacon is the Team Leader of the Robotics Research Team at Ocado Technology. He has been working with robots since his dissertation on Image Based Visual Servoing was judged the best control engineering masters degree dissertation in the UK by the Institute of Measurement and Control in 1985. He received his PhD from the Department of Artificial Intelligence at the University of Edinburgh in 1997 for showing how an inherently back-drivable robot could use sliding motion primitives to raise the level of programming abstraction in robotic assembly. He has also worked on error recovery, force-guided assembly, sensorless manipulation, skill acquisition, vision-guided handling, image-guided surgery and sheet metal folding. He has worked in the high volume environment of the automotive industry to the high precision environment of medical robotics. He is currently working on the handling and packing of groceries.
Although some groceries are regular shaped and rigid, many are such that their shapes and sizes are never the same twice, whilst also being susceptible to damage. Fruit and vegetables typify this class of objects. Ocado is interested in uncovering a principled way to handle such objects in an environment where they are only loosely constrained and is investigating soft manipulation as the vehicle to achieve this.