Amazon Picking Challenge (2017)
Why it matters
Since the mid 50’s, there has been an increase in robotics for industrial purposes. With the increase in control, precision, computational power and safety, robotics in an industrial setting has taken many leaps, however, there are still many problems that lie ahead before most processes are automated. One of the main issues around the automation of robots is that they need to be flexible and safe, that is, having a robot that is able to do multiple tasks safely.
This project aims to build a system that given a set of new objects in a random position, will be able to sort out each of the required items and safely manoeuvre them to different locations. Amazon, the global online store, has set a yearly competition. The Challenge combines object recognition, pose recognition, grasp planning, compliant manipulation, motion planning, task planning, task execution, and error detection and recovery. In addition to last years challenge, participants were also required to build their own storage system with most teams choosing to position totes on the ground.
At this year’s challenge in Japan, 16 teams from around the world competed, with the Australian Centre for Robotic Vision (ACRV), headquartered at QUT, winning the $80,000USD first prize.
Their solution, a Cartesian manipulator dubbed ‘Cartman’, moves along three axes, like a gantry crane, with a rotating gripper to allow item pick-up using either suction or a simple two-finger grip.
The Amazon Picking Challenge was designed to address a gap in the Amazon.com automated warehousing processes. Although Amazon can quickly package and ship millions of items to customers from a network of fulfilment centers (at one point recording up to 426 items per second), commercial technologies enabling robots to pick items and stow them in boxes in an unstructured environment are yet to be developed. The automation of this will have wide varying applications for Australian businesses, both with Amazon releasing a warehouse and a growth in warehousing businesses.
Cartman was used at the 2017 Amazon picking challenge, showing the results:
- Robust perception system (no need for high level of light control)
- Robust pick success (pulling a set of objects from a tote and storing them into other totes)
- 1st place out of 16 global teams.
Research into warehouse manipulators will continue over the coming years, with Cartman currently looking to be commercialised and spun-out into various industries.
Other Team Members
- Juxi Leitner - LYRO Robotics