It will require novel architectures both in software and hardware and will require AI algorithms compatible with physical, real time, robot operation. In terms of R&I advancement a paradigm shift is needed to remove silos between disciplines in order to weld together expertise and create a conceptual shift to reach the goals of next step autonomy for robotics.

Achieving this goal will require improvements in perception, awareness of the operating environment, the ability to anticipate and an improved understanding of the consequences of particular sequences of action on the working environment.

Proposals should address several of the following aspects of autonomy:

• Long-term, and where appropriate lifelong, autonomy of behaviour and energy (including frugality in terms of energy, lower environmental footprint, using new materials, designed to be recycled or easily repaired etc.)
• The autonomous adaptation of behaviours in dynamic environments.
• The development of robust and safe autonomy, including the development of risk averse systems or systems operating with low levels of communication or periods of communication denial.
• The use of high-level sources of information such as semantic information or externally held knowledge of the working environment, to improve autonomy.
• Mechanisms for advanced human interaction with systems capable of long-term autonomy.
• The impact of physical self-reconfiguration on autonomy
• The development of collective autonomy using multiple collaborative robots

This topic implements the co-programmed European Partnership on AI, data and robotics.