FELICE aspires to design the next generation assembly processes required to address urgent needs in manufacturing. It unites multidisciplinary research in robotics, AI, computer vision, data analytics, process optimization and ergonomics to deliver a modular platform capable to integrate and harmonize an array of autonomous and cognitive technologies aiming to increase the agility and productivity of an assembly production system and also ensure the safety and physical and mental well-being of human workers. To achieve these goals, technologies will be developed combining the accuracy and endurance of robots with the cognitive ability and flexibility of humans. Such flexible and configurable technologies will support future manufacturing assembly floors to become agile, and address Industry4.0 adaptation. We consider two layers of operation, control and processing, a local one introducing i) a collaborative assembly robot roaming the shop floor to assist workers in assembly tasks, and ii) adaptive workstations for optimizing worker-specific ergonomic aspects, and a global layer which will sense and operate upon the real world via an actionable digital replica of the physical assembly line. FELICE will i) implement perception and cognition capabilities allowing the system to build context-awareness, ii) advance human-robot collaboration in otherwise manual assembly lines, enabling robots to safely and ergonomically share tasks with humans, allowing the flexible reconfiguration of an assembly production process and, iii) realize a manufacturing digital twin, i.e. a virtual representation tightly coupled with production assets and the actual assembly process to enable the management of operating conditions, the simulation of the assembly process and the optimization of performance aspects and, iv) integrate developments in an industrial assembly line for agile production. FELICE has two pilot environments: one for technology validation and one related to car assembly.