Cracking the ‘Holy Grail’ of robotics research is one step closer thanks to a US$70,000 Amazon Research Award in recognition of the Australian Centre for Robotic Vision’s world-leading research into vision-guided robotic grasping and manipulation.
Centre Director Peter Corke and Research Fellow Jürgen ‘Juxi’ Leitner captured the global tech giant’s attention with a proposal to push the boundaries of grasping to ‘Grasping with Intent’. The pair, who led a Centre team to victory in the 2017 Amazon Robotics Challenge in Japan, join a coveted line-up of Amazon Research Award recipients, announced this week.
“Real-world manipulation remains one of the greatest challenges in robotics,” said Distinguished Professor Peter Corke. “So, it’s exciting and encouraging that Amazon is throwing its support behind our work in this field.”
Dr Leitner, who heads up the Centre’s Manipulation and Vision program described the Grasping with Intent project as ambitious and unique.
“While recent breakthroughs in deep learning have increased robotic grasping and manipulation capabilities, the progress has been limited to mainly picking up an object,” Dr Leitner said.
“Our focus moves from grasping into the realm of meaningful vision-guided manipulation. In other words, we want a robot to be able to seamlessly grasp an object ‘with intent’ so that it can usefully perform a task in the real world.
“Imagine a robot that can pick up a cup of tea or coffee, then pass it to you!”
Dr Leitner said the new project built on the Centre’s separate breakthrough in deep learning grasping techniques via development of an innovative real-time generative grasping code.
In 2018, Centre PhD Researchers Doug Morrison created an open-source GG-CNN network enabling robots to more accurately and quickly grasp moving objects in cluttered spaces. The breakthrough overcomes limitations of existing deep learning grasping techniques by avoiding discrete sampling of grasp candidates (unknown objects) and long computation times.
It focuses, instead, on one-to-one mapping from a depth image, predicting the quality and pose of grasps at every pixel. Fast enough for closed-loop control at up to 50Hz.
Dr Leitner said the Grasping with Intent project would add another layer of ‘intention mapping’ to the closed-loop GG-CNN network, opening access to task semantics.
“We envision, similar to pixel-wise segmentation of an object, pixel-wise labelling related to the task at hand.”
Communications Specialist, Australian Centre for Robotic Vision
P: +61 7 3138 4265 | M: +61 416 377 444 | E: firstname.lastname@example.org
About The Australian Centre for
The Australian Centre for Robotic Vision is an ARC Centre of Excellence, funded for $25.6 million over seven years to form the largest collaborative group of its kind generating internationally impactful science and new technologies that will transform important Australian industries and provide solutions to some of the hard challenges facing Australia and the globe. Formed in 2014, the Australian Centre for Robotic Vision is the world’s first research centre specialising in robotic vision. They are a group of researchers on a mission to develop new robotic vision technologies to expand the capabilities of robots. Their work will give robots the ability to see and understand for the sustainable well-being of people and the environments we live in. The Australian Centre for Robotic Vision has assembled an interdisciplinary research team from four leading Australian research universities: QUT, The University of Adelaide (UoA), The Australian National University (ANU), and Monash University as well as CSIRO’s Data61 and overseas universities and research organisations including INRIA Rennes Bretagne, Georgia Institute of Technology, Imperial College London, the Swiss Federal Institute of Technology Zurich, University of Toronto, and the University of Oxford.
Australian Centre for Robotic Vision
2 George Street Brisbane, 4001
+61 7 3138 7549