GROW-HBP

 

Human Brain Project models of scene understanding applied to General-purpose Robot for Object-retrieval in Warehouses

 

The overall goal of GROW is to develop the scene understanding technology of the Human Brain Project (HBP) and bring it to Technology Readiness Level 5 (TRL5) by integrating it into a system developed by CNR for the autonomous ’open-ended learning’ (OEL) robots. The system will be used to address the market need for the robot-based autonomous retrieval of sets of small objects from shelves in weakly-engineered warehouses. Scene understanding is critical in these scenarios as they present cluttered conditions that require disentangling the identity and features of a large number of objects that are highly variable in terms of visibility, shape, position, and orientation. The project will use simulations to illustrate the technology’s ability to account for a meaningful range of operational constraints. Key technological bricks will be ported to physical demonstrators, which will support discussions of take-up and exploitation with industrial partners.