Can robots pick single items efficiently? It’s an important question for e-tailers that need to pick a range of disparate items to fulfil an order. So it’s no surprise that Amazon put up $20,000 as first prize in a competition to find the best robot.
The Amazon Picking Challenge took place last week at the IEEE Robotics and Automation Society’s conference in Seattle and attracted 25 entries from universities and robotics specialists.
The challenge was to develop a robotic system that autonomously grasps several objects from a shelf, recreating the process that occurs in an Amazon warehouse when a client buys one or more products.
Points were awarded for a number of actions:
* Moving a target item from a multi-item shelf bin into the order bin
* Moving a target item from a double-item shelf bin into the order bin
* Moving a target item from a single-item shelf bin into the order bin
* Target Object Bonus: An added point bonus uniquely specified for each different object.
Penalties were assigned for other (unwanted) actions:
* Moving a non-target item out of a bin (and not replacing it)
* Damaging any item or packaging
* Dropping a target item from a height above 0.3 metres
The winner was the Robotics and Biology Laboratory (RBO) at the Technische Universität Berlin which got a total of 148 points.
A team from the Massachusetts Institute of Technology came second with 88 points. Third place was awarded to Team Grizzly of the company Dataspeed, with 35 points.
So, congratulations to Team RBO. But there is clearly some way to go before we see this technology in commercial use.
Apparently, given an order of 12 objects, it correctly picked 10 of them – a success rate of 83 per cent. So plenty of room for improvement.