"Cartman" takes top honors in Amazon Robotics Challenge

Think you can do automated fulfillment better than Amazon.com Inc.? The online retailing colossus wants you to prove it. Every year, Amazon hosts a robotics contest, challenging teams from universities around the globe to develop a robot that can identify, pick up, and transfer goods from warehouse totes into Amazon's cardboard boxes (and vice versa).

This year's Amazon Robotics Challenge drew 16 teams to Nagoya, Japan. For the competition, the robots raced against the clock to complete three rounds: removing target items from storage and placing them in boxes, taking items from totes to place them in storage, and finally stowing all items before picking selected inventory and placing it in boxes. Not only did the robots have to be swift; they also had to be accurate. Judges docked points from robots that left items unpicked, deposited goods in incorrect locations, or dropped or damaged items.

When the dust settled, a team from the Australian Centre for Robotic Vision (ACRV) had won the overall title, while Nanyang Technological University of Singapore won the picking task and MIT/Princeton of the U.S. won the stowing task. The winning entry—ACRV's "Cartman" Cartesian robot—can move along three axes at right angles to each other, like a gantry crane, and features a rotating gripper that allows the robot to pick up items using either suction or a simple two-finger grip, according to a press release from the center.

The competition itself is part of Amazon's push to further automate the order fulfillment process. In a statement on its website, the Seattle-based company acknowledged that while its current technology has removed much of the walking and searching from warehouse work, "commercially viable automated picking in unstructured environments still remains a difficult challenge." Amazon said it was looking "to strengthen the ties between the industrial and academic robotic communities and promote shared and open solutions to some of the big problems in unstructured automation."