British online supermarket Ocado has already made huge strides to automate its operations, using pick and pack robots in its warehouses and trialing self-driving delivery pods in London. Now it's looking to give its human maintenance technicians a helping hand with the SecondHands project, and has just unveiled its first robot assistant test prototype named ARMAR-6.

The EU-funded SecondHands project's purpose is to build a helper robot that will literally provide an extra pair of hands to maintenance and repair technicians working at Ocado's automated warehouses. This could be simply providing physical support for structures while the human techs are working, bringing equipment or materials when and where needed, taking over when the job calls for machine-level precision, or even predicting what's required to complete a task.

"The Ocado warehouses include state of the art automation and robotics systems and our technicians and operations engineers are tasked with ensuring that they function 24/7, 365 days a year," Ocado's Alexandru Voica told us. "These warehouses pick and pack millions of grocery items per day and deliver tens of thousands of orders to our customers in one-hour slots. The ultimate goal is for the robot to use machine learning and computer vision to scan its surrounding environment and identify tasks it could help with.

"So for example, if it observes a technician attempting to change a panel and requiring a set of tools, it will come and offer its assistance – either by holding the panel for the engineer or grabbing the various tools that are needed."

Creating such a collaborative robot assistant is very much a collaborative effort in itself, involving research partners from École Polytechnique Fédérale de Lausanne (EPFL), Karlsruhe Institute of Technology (KIT), Sapienza Università di Roma (SUR), and University College London (UCL).

EPFL, for example, has been working on the physical interactions between human and robot, including action skills learning, while KIT is responsible for the development of the robot itself, which covers its entire mechatronics, grasping and manipulation capabilities, software operating system and control, and the development of speech recognition systems. The natural language interface will allow the technician to converse with the robot in a similar way to people interacting with Amazon's Alexa or Google's Assistant.

Both SUR and UCL have worked on aspects of the robot's visual recognition smarts, with the former also developing its decision-making and task management abilities and task monitoring.

The ARMAR-6 prototype has now been delivered to Ocado's robotics research lab where it will be tested, evaluated and improved. Meanwhile each research partner will continue to work on individual components and system until the initiative comes to an end in 2020. After that, the teams will look into ways to get the robot helper into production and subsequent deployment.

You can see the ARMAR-6 prototype interacting with its test environment in the video below.

View gallery - 3 images