News › Forums › RAIN › General Discussion and Troubleshooting › Of Parallels, Sequencers and Timers
Tagged: Parallel Detect Sequencer
This topic contains 2 replies, has 2 voices, and was last updated by prime 2 years, 2 months ago.
January 23, 2023 at 7:44 pm #6604
So I’ve had a behavior tree ‘SimplePursuit’.
—->AICategorizeTargets (Custom action, detect and dump targets in to friendly/hostile/neutral lists)
—->AIChooseTarget (Custom Action, choose a target from the target list)
—->move (move toward the target using a custom motor).
This works fine. However, I don’t really want the detect/categorizing running every frame, that’s quite overkill.
After a bit of research, it seemed like the following should work:
——>Timer (1 second delay)
All items set to ‘repeat never’
I’d have expected the Sequencer and move to execute in parallel. Instead it simply repeats
the items in the sequencer once per second as I’d like, but the move node is never called. If I
drag the gameObject within detection range of a target, then the move node executes… but only
the move node. Once it has a valid target and starts executing, it seems the sequencer never happens again.
Sequencer set to ‘repeat forever’:
It sits and detects. The move node never happens.
Sequencer ‘repeat never’, any node under the sequencer ‘repeat forever’:
That node (AICategorizeTargets or AICHooseTarget) repeats forever, no other node executes.
Clearly this isn’t behaving at all as I expect. It seems like the parallel node won’t handle
the sequencer parallel to the move? Am I approaching this fundamentally wrongly?January 23, 2023 at 8:20 pm #6606
FYI, I have figured out a combination that works:
——>Timer (1 second delay)
where the Sequencer is ‘repeat forever’ and the move node is ‘repeat forever’. I believe this was not working because,
according to the wiki:
Child nodes in a Parallel will execute until Success or Failure, but will not be automatically restarted by the Parallel. Once a child node has completed, the Parallel may continue executing other child nodes until a Succeed or Fail condition is met.
Thus both the move and the sequencer need to be ‘repeat forever’ or they are never run after the first pass through the parallel. Is this correct? I suppose I can see the logic of this. Each item in the parallel needs to have completed before any of them can have a second execution, hence the ‘not automatically restarted’, which is quite a different behavior than the other nodes.
January 25, 2023 at 10:53 pm #6629
- This reply was modified 2 years, 2 months ago by DrHeinous. Reason: Added a bit of reasoning about parallels
The problem is likely that the first time the move node runs, it does not have a valid target. That’s because choose target may not have run (it has at least 1 second delay). That could cause the move node to fail. If the move node fails, then your parallel could be failing - before the choose target ever happens. Repeat…
Once you set your move node to repeat forever, it no longer returned failure, which allowed the sequencer time to complete.
You must be logged in to reply to this topic.