Continuous dialog without wake word

What would be the best way to have rhasspy work in a continuous dialogue without a wake word.
For example i want to wake a robot, dictate commands continously, then close the conversation
Sample:

  1. run a program
  2. start text to speech
  3. “Robot wake up” --> Robot: “OK”
  4. “Left 30 degrees”, “Stop” … “Raise left arm”… “Come here”… etc.
  5. “Robot go to sleep” (no more text to speech) all dialogue ended

I have been able to do this with the system as is, but its sluggish, and it’s redundant, slow and wasteful to use the wake word for every command.
A bonus would be to have a closed dialogue in the midst of the commands. For example “Robot pick up the closest object” Robot: “is this what you want?” , “No”, Robot: “Ok, i will try another” (bad example but hopefully you get the point).

I know all the parts i need are right there in front of me, but im not sure of the correct sequences and messages, and if i could, i prefer to use hermes mqtt.

Any ideas would be appreciated.

A lot of the ‘wake word’ are key word systems for lightweight low vocabulary recognition and guess you would have to dive into training up your choice of KWS and use that in some sort of multiple key word system.
I think the majority of KWS employed with Rhasspy will work that way, just not implemented that way.