search

srobo / competition-simulator

A simulator for Student Robotics Virtual Competitions
https://studentrobotics.org/docs/simulator/
MIT License
9 stars 2 forks source link

Clicking 'Reload World' does not restart the robot #351

Closed Adimote closed 1 year ago

Adimote commented 1 year ago

if you click the 'Reload World' icon: image

It restarts the simulation, but it doesn't run the robot code

The log messages after a reload look like this:

INFO: sr_controller: Starting controller: python.exe -u sr_controller.py 0
INFO: sr_controller: Starting controller: python.exe -u sr_controller.py 1
INFO: sr_controller: Starting controller: python.exe -u sr_controller.py 2
INFO: sr_controller: Starting controller: python.exe -u sr_controller.py 3
INFO: competition_supervisor: Starting controller: python.exe -u competition_supervisor.py
INFO: lighting_controller: Starting controller: python.exe -u lighting_controller.py
No robot controller found for zone 1
INFO: 'sr_controller' controller exited successfully.
No robot controller found for zone 2
INFO: 'sr_controller' controller exited successfully.
No robot controller found for zone 3
INFO: 'sr_controller' controller exited successfully.
Development mode, exiting competition supervisor
INFO: 'competition_supervisor' controller exited successfully.
PeterJCLaw commented 1 year ago

It looks like you've ended up paused, were you paused when you clicked restart? For me if the simulation is running then when I click restart the simulation starts immediately.

I'm also not sure this is something we've done -- I think you're just experiencing webots behaviour.

PeterJCLaw commented 1 year ago

We dug into this a bit yesterday. It turns out to be to do with the lighting controller constructing a Supervisor (robot) instance after the restart. That call never returned meanwhile the user-code doing a (Webots) Robot.step also never returned. Together this seemed to deadlock the simulation. This suggests that this is a bug in Webots, however the latest nightlies of Webots don't have this issue so it may be fixed already.

PeterJCLaw commented 1 year ago

For completeness: it looks like this was probably https://github.com/cyberbotics/webots/issues/5279.

PeterJCLaw commented 1 year ago

I'm actually not completely certain that that situation aligns with what we saw, given that in our case there wasn't a crash, however the symptoms are otherwise very similar and the fact that the issue we observed seems fixed in the nightlies does suggest that it's been fixed around the same time.