No idea what you are stuck on.
Easiest way.
inside locomotion add another state machine.
create a boolean called alternate idle.
Create an integer called idleID
Inside the new state machine add a state called “start” put a 1 frame copy of the end pose of idle. It’s literally only used as “backup”.
From this state machine create many others. To transition into it make use of the idleID int AND the boolean (2 is better then 1 for error prevention).
inside those states, you plug in the idle animation and disable loop. Possibly disable blend time or set to .01
Then you create a return pathway with “automatic” which should simply go back once the animation is done.
Make sure you also have a 2 way on the start state machine to entry.
Technically that’s all.
You then need to figure out the best way to pick a random in range.
You so this using the notify so its selected at the end of the idle before the transition happens.
I would keep the notify 2 frames before the end.
in normal conditions the notify will cause the transition to happen just before the last frame, which is why that’s used in start. And the transition from start to the new idle is probably less than frame, so you potentially do skip the end frame.
The worse case scenario possible is that the frames go out of sync and the automated blend time system picks up the slack masking any errors in the transitions.