Team Delft has qualified as one of the 16 finalist teams for the Amazon Picking Challenge 2016. The team is a joint effort of the startup Delft Robotics (Kanter van Deurzen a.o.) and TU Delft Robotics Institute (Carlos Hernandez Corbato a.o.), supported by the RoboValley initiative (www.robovalley.com)
The goal of the Amazon challenge is “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." In order to spur the advancement of these fundamental technologies, there will be two parallel competitions: the Pick Task, and the Stow Task. For the Pick Task, target items for an Amazon order have to be removed from a standard shelf in Amazon warehouses and placed into a tote. The Stow Task requires the reverse: target items have to be taken from a tote and stowed into the bins of the shelf. These tasks involve challenges in object recognition, grasping, dexterous manipulation, and motion planning.
Since January, Team Delft has been developing an industrial grade robotic system for the challenge. It involves a 7 degree-of-freedom Motoman robot mounted on a rail, courtesy of Yaskawa (sponsor). Ensenso cameras from sponsor Imagining Development Systems will feed high quality 3D images to a vision pipeline for object recognition and localization using Deep Learning techniques. The team is fully committed to the ROS-Industrial initiative. ROS and ROS-Industrial components for motion planning, robot control, grasping, and PointCloud processing will be integrated into a fault-tolerant control architecture for the robot.