The Robot Vision research group from the University of Alberta is working on image guided motion control of robot arms and hands. This team is implementing processes that allow the robot to learn from humans by means of observation, gestures and dialog so that future robot systems will be able to work with humans even in unstructured environments. The goal is for the robot, with this acquired knowledge, to be in a position to grip various everyday objects, workpieces and components and to sort them independently, even if new, unknown objects are included.

"We are thinking the fair is like a window to the future, a place where you can present and discover real innovations. For us, it is a great opportunity to build a bridge between research and industry."

The project brings together experts from the Istituto Italiano di Tecnologia in the fields of dynamic human modeling, image processing and robot interaction control. Its objective is to deploy robots to support and guide humans in laborious industrial processes in order to ease the strain on them and prevent injury. When collaborating with a human, the robot monitors the dynamic performance and mental condition of its counterpart.

"The idea is to make robots collaborate with humans - not to avoid them. This requires a lot of data fusion and is the biggest challenge we have."

The team from the Institute for Computational Design and Construction at the University of Stuttgart combines expertise in algorithmic geometry and the development of robot-based material processing systems. In this project, it aims to provide small and medium-sized companies with access to robot-assisted methods of work. The concept comprises a collaborative robot workbench with an augmented reality interface. The project will demonstrate a woodworking scenario in which a robot assists a human.

"Existing robotic interfaces and workflows require extensive expert knowledge and experience. We believe that they should radically change. With the help of Augmented Reality, non-expert users should be able to engage with robots as well."

The employees of Draper, a non-profit R&D organization, the Robot Locomotion Group at MIT and the Agile Robotics Lab in Harvard, aim to show that robots can orient themselves and execute tasks in unstructured environments. For this, the team uses neural networks to determine the positions and interactive dynamics of objects in the environment. The team’s developments are demonstrated by means of a maintenance task that is typically carried out by a human. 

"If a robot is working next to a person at a cluttered workbench or in a crowded kitchen, it needs to be able to effectively recognize objects and to manipulate them – without really knowing anything about them beforehand.

Four researchers from the University of Pennsylvania combine their expertise from various robotics disciplines ranging from image processing and manipulation planning to system integration. The team is tackling the challenges facing a robotic system working with containers filled with liquids and used for exact dosing, for example in the pharmaceutical industry. The robot is controlled using sensors on the robot arm and two stereo cameras.

"We believe that our project really embodies the goal of KUKA: To have robot team mates that can work effectively and safely hand in hand with human colleagues."