Congratulations to the team which has recently won the 2017 Amazon Robotics Challenge scoring an impressive 272 points after a finals round in which both picking and stowing tasks were accomplished without blemish (the runner-up team scored 235). The team’s share of the prize pool of US$250,000 will be around US$100,000.

The Amazon Robotics Challenge is an open competition for Industry and Universities and Industry groups to propose solutions to the problem of picking and stowing a large variety of items into “totes” or baskets. This remains a manual task and is a barrier to full automation of warehouses for Amazon and other storage and distribution facility operators which already use robots to move shelves and items around the warehouse floor.

For the last 3 years Amazon has run this competition. This year the final was held in Nagoya, Japan in association with the Robocup Competition/Symposium.  Amazon organisers selected 16 international teams (including a joint entry from MIT-Princeton and one from CMU, as well as teams from Japan, Singapore and Germany — UBonn, last year’s runner-up) following a competitive bidding/application process.

The Australian Centre of Robotic Vision used this competition as a means to showcase our world-leading robotic perception research and technology. The team was built around a core of engineers and researchers from QUT and University of Adelaide nodes of the Centre (with some participation from the ANU node also) and featured a custom-designed and byuilt cartesian robot, “CartMan” — which was transported in parts from Brisbane to Nagoya in team members’ carry-on luggage and reassembled on site — and the Centre’s world-leading deep learning segmentation algorithms, developed in Adelaide, for finding and identifying the objects.

Centre Deputy Director and Adelaide Node leader, Prof Ian Reid, said of the victory: “This a great demonstration of the quality of Australian robotics and computer vision research. It showcases the our world-leading expertise in deep learning for robotic vision, and is a great example of the collaboration happening between our institutions.

This could not have been achieved without the robotics and mechatronics expertise in QUT and the computer vision and perception expertise in Adelaide and ANU coming together to create a robotic vision system to solve a real-world task in an unstructured environment.  This is what our Centre is about: bringing perception and robotics together to create robots that see.  I want to congratulate the team for their achievement, which is just reward for their dedication and hard work over the last few months.”