I would suggest as a starting point, that you get some sheets of clean paper and structure what you want to happen. Your phrasing on the question is vague, at best, so you need to clarify things in your plan down to boolean (
true/false) values for a computer to understand.
You want your AI to "wander around". Does that mean they wander in random directions? What would be a minimal length of a directon (It can look meaningless of the scale of your random walk is gameworld millimetres rather than metres, for example).
Then you want your AI to "interact" with objects. Which objects, what is the criteria for interactions - proximity? time of day? mood?
You need to set a clear process tree and for each subroutine, such as "pick up desk lamp and wave it around for 5 seconds", you need to set what criteria needs to be forfilled by your AI before this event is begun, such as
- "is AI within 5m of a desk lamp?",
- "is AI script state marked as "bored" " ,
- "is AI marked as Awake" (rather than sleepwalking!)
If these criteria are all true, then you can start the "play with desk lamp" subroutine,
Overall I would suggest you need to really clarify the behaviours and routines you want to create, and what would cause them and enable them. Then once all these "factors" are set out clearly in your mind and in your plan, you can work on coding each one, one by one, until you have a full behaviour setup.