MERGING will deliver a turnkey robotic solution to automate handling of flexible and fragile objects. Our ambition is to provide manufacturers with an end-to-end solution to automate the handling of soft objects. The solution will consist of a fingered gripper equipped with an electro-adhesive skin that conforms to the objects to handle, even delicate fabrics or components, without any damage. Electro-adhesion increases the direct gripping forces and thus allows greatly reduced clamping forces. Our solution includes perception and supervision functions to adapt the system's behaviour in real time to the execution conditions and for robot system programming accessible to non-specialists. Our main motivations are to build a versatile, easy-to-use and low-cost system. To demonstrate this and the possibilities of scaling up, we will design our system using proven laboratory technologies (TRL 4), then carry out proof of concept in realistic environments (TRL 6) in three different applications and sectors: fabric handling for lingerie manufacturing, technical fiber handling for composite bus panel manufacturing, plastic bag handling for the food industry. MERGING is a four-year project involving the entire value chain of soft object gripping automation. The CEA leads the project and provides the robotic technologies in collaboration with EPFL (electrical adhesion), AIMEN (perception), LMS (supervision), and SHADOW then OMNIGRASP (capture) and IPC (materials). CEA, EPFL, AIMEN and LMS will respectively use AI technologies for robotic demonstration programming, electro-adhesive skin control, perception and control of the complete system. SELMARK, VDL and THIMONNIER introduce cases of use of fabric, composite and polymer handling respectively. CASP and OPTEAMUM are respectively responsible for the software and hardware integration and the exploitation of the results. OMNIGRASP will release the pliers with electro-adhesive skin after the end of the project.

HARTU will provide the necessary tools to implement three basic stages of parts handling, i.e., grasping, assembly and releasing. These tools aim to address the key challenges for creating handling applications by means of innovative technical approaches: (1) self-supervised grasp and release planning policies identification and control; (2) to learn and control contact-rich assembly skills from human demonstrations; (3) to develop an AI-based multi-modal perception for visual-servoing and continuous monitoring in handling operations, supported by virtual and continuous learning; and (4) to develop versatile and dexterous soft grippers with electro-active fingertips. HARTU is an industry-driven research project that includes the deployment of technologies in 5 industrially relevant scenarios: 4 manufacturing lines in different sectors: automotive, household appliances, hand tool manufacturing and food processing, and 1 in logistics to validate that the technologies can be transferred to other sectors. They offer an enormous variability of products in terms of shape, material and sizes. This approach contributes to the (1) industrial objective of HARTU, i.e., to increase the flexibility, reconfigurability and efficiency of manufacturing lines through easy to integrate and configure, safe and reliable handling systems, supported by tools for application development and a reference architecture. It also contributes to the (2) social objective: to contribute to user acceptance and adoption, identification of skills development needs and compliance with the liability/legal and ethic aspects.