Increasing memory consumption iteratively creating and deleting camera nodes

[msminirobot]

trafficstars

Describe the Bug In an evolutionary robotics setting, I create new robot nodes iteratively and delete the old ones using a supervisor node. Over time, the memory consumption increases up to very high levels (> 30 GB). Isolating the other elements of my robots, I notice the memory increase only when enabling a camera device

camera = robot.getDevice('rgb_camera')
camera.enable(params.TIME_STEP)

Steps to Reproduce

1. Create robot node with camera using a supervisor node
2. Delete the robot node with the supervisor
3. Create a new supervisor node
4. Repeat steps 1-3 iteratively, you will see the memory usage of Webots increasing over time

Expected behavior Memory consumption does not increase continuously.

System

• Operating System: macOS 15.0.1
• Python 3.12

Additional context Add any other context about the problem here.

[austingreisman]

Sounds like a memory leak (I'm not a contributor, but use Webots). Is it possible for you to use the camera.disable() method before deconstruction to see if that "fixes" the issue?

[msminirobot]

Yes, this is a memory leak. However, I don't know which part of the camera node it exactly causes.

Using disable, deleting camera object and/or robot node doesn't help. I tried all of those combinations and my guess is that there is something not properly removed by the destructor.