SMART-E is a Doctoral Training Network, funded by the EU Commission’s FP7 Marie Curie Programme , with the aim to prepare the next generation of leading experts in advanced robotics to secure a sustainable manufacturing sector in Europe.

The network involves the collaboration of eight European institutions – the University of Salford, University of Zurich, Scuola Superiore Sant'Anna, Italian Institute of Technology, Technical University of Munich, FESTO, Airbus and Advanced Manufacturing Research Centre at the University of Sheffield.

The network is bringing together leading experts in the field of embodied intelligence, compliant, soft robotics, advanced manufacturing, industrial robotics and automation, smart materials and machine learning in Europe.

• Designing and delivering a globally leading, sustainable doctoral training programme;
• Providing the participants of the programme with complementary business, leadership and interpersonal skills; exposure to different working cultures in academic and business sectors internationally.
• Devising innovative solutions for industrial applications in Advanced Robotics and Intelligent automation for sustainable manufacturing in 3 broad topic areas i) Dexterous, Soft and Compliant Robotics in Manufacturing; ii) Reconfigurable and Logistics Robotics; and iii) Safety and Human Robot interaction and cooperation

TU München leads the topic area "Safety and human robot interaction and corporation" as well as the quality management of the overall project. The following sub-projects are directly executed by TU München:

• New control methods for musculoskeletal-based robot manipulators
• Safety verification of human-robot co-working
• Modular and bionic-inspired light-weight robot

In classical industry automation robots are separated from human workers for safety reasons. The disadvantage of this setting is that the robots require huge and expensive safety cells and that due to the separate workspace the manufacturing becomes unflexible. In order to be competitive in flexible manufacturing with products produced in small to medium quantities, humans require assistance of robots. Sharing the workspace of robots with humans has many advantages since complementary abilities can be usefully combined. For instance, cognitive capabilities of humans can be joined with the precision and power of robots. The problem, however, is that this set-up is no longer safe. Two approaches to achieve safety are pursued:

• Building of soft, flexible robots that are safe by default due to their small mass, soft body shell and flexible joints. In order to achieve those properties, musculoskeletal-based robot manipulators are designed and built. The challenge of musculoskeletal-based robot manipulators is that they are much harder to control than classical serial robots with rigid links. New nonlinear control methods are developed to overcome the difficulties.
• Researching new verification methods to certify that the robot never hits a human. This should be done with formal methods that make it possible to mathematically prove that the system is safe. Formal verification of safety in human-robot interaction is an open topic and provides great opportunities for ground-breaking research results. The challenge is that the safety verification has to be performed while the robot is in operation, which requires fast and smart algorithms to prove freedom of collisions. It is expected that the techniques developed for this problem are beneficial for other safety-critical systems such as automated vehicles, which might also be used in future production sites.