Personality Generator Canceling Long Scripts

Yes - what you’re seeing is expected behavior for Personality Generator. It runs one personality script at a time, and when the next randomized interval hits, it cancels the currently running script and starts the new one. So a long routine like your 20-30 second sweep will get interrupted unless you design around that behavior.

Recommended pattern

Option 1: Keep Personality scripts short and non-blocking

This is the simplest and most reliable approach.

Instead of one long script that "does" the whole sweep, make the Personality Generator trigger small actions only:

• move head a little
• speak a short phrase
• blink LEDs
• change pose briefly

Then let the next trigger handle the next small step.

This works well because Personality Generator is meant for spontaneous, short behaviors, not long routines.

Option 2: Use a state machine with short scripts

If you want the bot to appear to do a long routine, break it into phases and store the phase in a global variable.

For example:

• head_state = 0 start scan
• head_state = 1 continue scan
• head_state = 2 finish scan

Each Personality Generator trigger runs only one phase, then exits quickly.

That avoids cancellation issues because each script is short.

Option 3: Have Personality Generator call a Script Collection script that only starts if idle

If you want to keep using Script Collection, the key is:

• Personality Generator should trigger a script that checks whether another action is already active
• If active, it should immediately exit
• If not active, it should start a short action and set a state flag

But note: if the Personality Generator itself fires again, it can still cancel the current running script if that same script is still executing. So this only helps if the triggered script is very short.

What won’t solve it by itself

These approaches usually won’t prevent cancellation:

• just adding $busy = 1
• putting the routine in Script Collection and calling ScriptStart
• widening the interval a lot

Why? Because the Personality Generator doesn’t wait on your $busy flag - it simply starts a new script when its timer fires, and the currently running personality script gets cancelled.

Best practical workaround

If you want long ambient behavior, use Personality Generator only as the "starter", not the executor.

Example pattern:

1. Personality Generator runs a tiny script:

   • if not busy, set a mode variable like $lookState = 1
   • start a separate worker script using ControlCommand("Script Collection", "ScriptStart", "LookAroundWorker")

2. The worker script should be designed to:

   • run in short chunks
   • save its state
   • exit quickly, or use repeated short calls rather than one long blocking loop

If you need a long-running routine, it’s better to trigger it from a different control than Personality Generator, or convert it into a repeatable state machine.

Example pattern

Personality Generator action

if (getVar("$busy") == 0) {
  setVar("$busy", 1);
  setVar("$lookState", 1);
  controlCommand("Script Collection", "ScriptStart", "LookAroundWorker");
}

Worker script

if (getVar("$lookState") != 1) return;

Servo.setPosition(d5, 70);
Servo.setPosition(d6, 90);
sleep(500);

Servo.setPosition(d5, 50);
sleep(500);

Servo.setPosition(d6, 120);
sleep(500);

Audio.say("Scanning complete");

setVar("$lookState", 0);
setVar("$busy", 0);

This is still not a perfect "queue," but it keeps the Personality Generator from being responsible for the entire long routine.

Important note about speech

If you’re using speech in the same long script, that also increases the chance you’ll see interruption. A better pattern is:

• keep speech phrases short
• trigger speech as its own short action
• avoid long scripts that mix movement loops + speech + delays

Relevant docs

• Personality Generator: https://synthiam.com/Support/Skills/Scripting/Personality-Generator?id=16084
• Script Collection: https://synthiam.com/Support/Skills/Scripting/Script-Collection?id=16045
• ControlCommand(): https://synthiam.com/Support/Programming/control-command
• Create custom skills if you need a true queued behavior system: https://synthiam.com/Support/Create-Robot-Skill/Overview

My recommendation for your LattePanda/Leonardo setup

Since you’re already on ARC with Windows and the onboard Leonardo, I’d recommend:

• keep Personality Generator actions under 1-3 seconds
• use a state variable for multi-step head scanning
• let the scan continue across multiple triggers rather than one blocking loop

If you want, I can help you rewrite your current look_around routine into a non-blocking state machine in JavaScript or Python for ARC, which is usually the cleanest way to do this.