Team

• Research: Manufacturing Technology Centre (MTC)

• Research: Loughborough University (LU)

• Research: University of Birmingham (UoB)

• End User: Airbus Operations GmbH (Airbus)

Joint Robot-Human Logistics and Assembly in Aerospace

The scenario is an assembly operation on the inside or outside of an airframe. The robot will pick and carry a large number of fixing elements or assembly parts and the relevant tools from a storage area and bring them to a worker at the airframe assembly. The robot and the human will then work together to apply the fixing elements and parts. The key steps in the scenario are:

• Identifying and grasping required tools and materials from a work-surface.

• Placing the tools and materials on the mobile platform in a safe configuration for transport.

• Navigating to the worker, and waiting alongside.

• Grasping materials or a tool taking into account wrenches to be applied in its operation.

• Positioning the material or presenting the tool for co-manipulation by the human, who will then apply it to each fixing to perform the task.

• The robot will learn the human applied forces, and assist the human by adapting the end effector forces over time (i.e. in a gravity compensated mode).

• The robot will interpret foot motion in order to follow the human as they walk along the airframe to enable continued fixing.

This scenario covers a variety of important human-robot collaboration skills in a shop floor logistics and manufacturing setting. Our contention is that to be useful in many practical settings it is as important to allow the robot to be used as intelligent tool guided by the human, as it is for the robot to operate fully autonomously, e.g., for picking and moving parts. This proposal is predicated on the notion that to achieve fluid interaction much of the technical challenge lies in the robot knowing when to switch from one mode to the next on the basis of interpreting the human’s intentions from sensory input. The key scientific and technical challenges that must be met which go beyond the state of the art are:

• Flexible Grasping - Grasping and fetching of parts and tools of possible novel shape. We will build on our recent breakthrough work on this.

• Force based Human-Robot interaction - The robot works as a multiplier of the human’s own abilities, co-applying forces at the end effector in the desired way.

• Adaptive Interaction and Machine Learning - Learning will be used to allow the human to train the robot over time to apply assistive forces optimally. In addition the robot must interpret the human’s body motions to correctly follow the human.