Digital twin (DT) is popular in industry to accelerate development and manufacturing of products or production. Interestingly, we could also use digital twin during system operation. This is called executable digital twin (xDT): a self-contained virtual representation that models physical behaviors and combines with physical data to provide augmented information.
Recently, we have developed an application of xDT for automatic vehicle controller tuning during runtime. The main idea is to implement multiple xDT of high-fidelity vehicle dynamics models in parallel with physical vehicle. The xDT information is fused with physical data to learn and tune an optimal controller, leading to higher control performance.
If interested, please check here the paper presented by my PhD, Jean Pierre Allamaa in the IFAC Automotive Control Conference in Ohio last month: https://lnkd.in/eQfgwsUf.
This is the collaboration work in the EU ELO-X project, together with Herman Van der Auweraer, Panagiotis (Panos) Patrinos and myself.
Also relevant, if you want to know how we develop such xDT model, please see here a short demo video: https://lnkd.in/ecGrBUcp