Pico W : Script not depending on UART or REPL but needs me to press several times RETURN key on REPL (mpremote) to continue execution

[mchobby]

I have a really strange issue with my Pico W running on MicroPython v1.19.1 .
The project is a port of the robot zumo (for pololu) to make it running with Pico/Pico W under MicroPython.
Here a picture from the micropython-zumo-robot GitHub repository.
The test_readline3.py code copied to main.py is quite simple an read the value from line tracker sensor (I will write details later) (no boot.py file in this setup).

from zumoshield import ZumoShield
import time
z = ZumoShield()

z.play_2tones()
z.button.waitForButton()

# Reload calibration data
z.ir.calibrationOn.load_json( '{"maximum": [2000, 1775, 1475, 1172, 1500, 2000], "minimum": [303, 303, 303, 303, 303, 304], "initialized": true}' )
z.ir.calibrationOff.load_json( '{"maximum": null, "minimum": null, "initialized": false}' )

counter = 0
print("Move the Zumo over the line to read line position.")
while True:
	# With the Zumo blade going forward
	#   Value from 500 to 4500 : line is placed from left to right
	#   value 2500 : line centered on the zumo
	#   value 0 : line exceed on the left
	#   value 5000 : line exceed on the right
	position = z.ir.readLineBlack()
	counter += 1
	if (counter % 100)==0:
		print( '%6i: Line position: %i' % (counter,position) )
	if (counter % 1000)==0:
		z.play_blip()

I have no memory issue and no stack issue (carefully checked, also controlled with mem_info() ). No memory fragmentation.
It often got stuck at one moment or the other. After I applied the RETURN key solution the script can run up to 382K iteration without issue.
A day (by hazard), I discovered than pressing the RETURN key in REPL did make the script executing one line forward (because IR LED activation did Light on)... after several RETURN keys pressed in REPL the script suddenly restart running "forever".

The way it work, the QTR sensor from Pololu requires the Pin to be set as OUTPUT for charging capacitor then switch back to input an measure the decay time. So, the Pin mode is constantly changed while using the sensor.

If REPL is not available (because not connected to PC) then the script waits forever.

Could the VM get stuck (on hold) for strange reason ?
This issue driving me completely crazy.

[mchobby]

I moved the code to a Pico (no wireless) and looks to work pretty well.