Robots and machines are continually exposed to change – internal variations such as component wear and tear, and external variations such as fluctuating environmental factors.

Conventional control systems are not equipped to deal with those variations. They are inflexible, optimised for certainty using digital models that do not transition well to real-world operating conditions. The result rarely meets the expectation: suboptimal performance, substandard quality, with a direct impact on productivity and sustainability.

Al brings unprecedented solutions to engineering problems. But Al too has limitations. Existing Al technologies struggle with the transition into physical hardware – what is known as the «reality gap». In order to deal with the uncertainty, they require huge amounts of data and computational expense.