Using model-based AI, the technology learns control rules automatically to reduce the cost and time of system development. A state space model of the particular system's dynamics is learned through repeated trial and error, based on which the technology automatically learns control rules needed to reach pre-defined goals. Manual program development by humans to teach proper control actions becomes unnecessary, which helps to reduce the time and cost of developing control programs.

Going forward, Mitsubishi Electric will continue to develop the technology by adding increased AI capability to improve the speed and accuracy of autonomous learning. It will also employ 3D simulations of the controlled equipment to significantly decrease the time required to learn control methods.

Mitsubishi Electric Corporation published this content on 14 February 2018 and is solely responsible for the information contained herein.
Distributed by Public, unedited and unaltered, on 14 February 2018 04:05:02 UTC.

Original documenthttp://www.mitsubishielectric.com/news/2018/0214-b.html

Public permalinkhttp://www.publicnow.com/view/5FCBC36B89BCB3FEFECCDC5C47B5D0A7EFFE8D09