Control at What Cost? One-Shot Real-Time Dual Inverse Optimal Control. This project will develop dual inverse optimal control to enable engineered systems to actively impute (learn or infer) and optim

Description

Control at What Cost? One-Shot Real-Time Dual Inverse Optimal Control. This project will develop dual inverse optimal control to enable engineered systems to actively impute (learn or infer) and optimise costs associated with control actions in uncertain dynamic environments. Dual inverse optimal control will be developed by introducing novel measures of cost imputability in optimal control, and by introducing real-time schemes that actively select control actions with the dual objectives of simultaneously improving imputability whilst optimising uncertain costs. The outcomes of this project will enable vehicles to actively infer drag and friction costs; robots to actively learn from humans by requesting demonstrations; and appliances that optimise their electricity consumption via imputing market-based costs.. Scheme: Discovery Projects. Field: 4007 - Control Engineering, Mechatronics and Robotics. Lead: Prof Iman Shames