Machine Learning and Deep Learning in Control of Machines

I think you said many interesting points but I will ask the first last one. Not every problem has to be solved by machine learning or it depends and sometimes it should be for example if you can't use classical control. Do you think that something frustrates you that we follow the trend in research? It certainly will be, it seems to be perceived as being cool in research but it's very inefficient. If it was a robot involved I'd probably figure out the kinematics and write down the forward kinematic equation at university of Macquarie.