Is It Possible to Combine a Data-Based Model and a Physical Model?

Physically informed neural networks can embed the knowledge of physical laws that govern a data set in the learning process. The best possible compromise is really to combine both worlds, like the data-based model and the physical model. And for example, if frictions are unknown, just let the AI learn this particular component of the system. Other components which are well known can be modeled physically, and then let the different models interact with each other.