it uses speed control for an induction machine. You can view how the controls are implemented. As far as tuning goes, you should typically tune the fastest loops first, so that would be your iq and id in this case. The particular example actually uses 4 pi controllers instead of 3.
To make things easier with tuning, if you implement v/hz control, you can see what realistic values for id and iq are for your machine under load to help with your initial tuning before you have all the loops working. (you might already know this, in which case you can ignore this)
Also to help, while your id and iq pi controller gain values shouldn't be the exact same to get the best performance, they will typically be in the same order of magnitude, so you could start tuning your current loops by keeping the gains identical until you get something working.
There is also the option of using classical control theory (or our tool, simulink design optimization) to help tune your controllers: Make sure to use average inverter models when doing classical control so you can use time and frequency based techniques. https://www.mathworks.com/help/slcontrol/graphical-and-automated-control-design.html?s_tid=CRUX_lftnav This will be harder than tuning by hand the first few times, but in the long run will allow you to design better controllers faster with more control over the response dynamics and stability margins.