How to run the simulation on local machine to test my code

[rua0ra1]

Hi,
My name is Ram and I am participating in the Humanoid Competition. I am quite new to the we-bot simulation. I am wondering how can I run the simulation on my local machine to test my code. Currently, I am pushing my changes to Github and analysing the simulation result in webot competition page. I feel its not a idle way to test my code. I really appreciate it if you can help with this issue.

[omichel]

You should download and install Webots R2023a on your local computer. I would recommend that you install R2023a-rev1 from a recent nightly build as it fixes some problems with Python controllers. Then, you can git clone on your local computer your repository with your code for the competition. Finally, you can open the wrestling.wbt world file from the worlds folder of your repository cloned on your local computer. You will then be able to run the simulation locally, make changes to your robot controller, test them locally and git commit them to the main branch of your repository. They will be run against the other participants only after you git push them on GitHub.

Note: if you are not familiar with git, I would recommend you to follow some introductory tutorial about it.

[omichel]

Unfortunately, there is no motion editor any more.

[pemile29]

Bad news !

It seems to me an indispensable tool, no?

[omichel]

Not really.

1. You can edit these motion files manually (they are plain text).
2. You can use the motion library to make the robot walk.

[kalashjain23]

When running locally, you should install the python modules on your local system, e.g., pip install numpy, etc.

I've installed all the required modules for my robot, but it still shows the same error:

import numpy as np
ModuleNotFoundError: No module named 'numpy'

When I'm running pip install numpy:
Requirement already satisfied: numpy in /usr/lib/python3/dist-packages (1.21.5)

Can you guide me through this one?

[Jean-Eudes-le-retour]

This doesn't seem like a Webots problem.

It looks like you might have multiple versions of python installed on your system, have a look at these links for pointers:
numpy/numpy#15424
https://www.quora.com/What-is-Python-m-pip

[pemile29]

The motion file library is still very limited!

I used it to write a simple spinning robot (which was first until 30 minutes ago) but I guess it won't be enough for long !

Editing motion files by hand doesn't seem easy to me !

PS: Is it possible to use the GPS to avoid going out of the ring ?

[pemile29]

I can't imagine how to use odometry that would take into account crashes and falls !

Using vision to locate oneself inside the ring seems beyond my capabilities with opencv :-/

Anyway, thanks for your help

[omichel]

Just some clues: using the second camera of the robot, which looks down, it might not be super difficult to locate on the camera image the red line delimiting the ring and to estimate how far the robot is from this line by measuring the apparent width of the line...

[pemile29]

Right, thank you for motivating me, although my reflex in front of opencv is to use TurnLeft180.motion !

[pemile29]

One more novice question, in the "Eve" demo what does the camera.send_to_robot_window() function do ?

And how can i display the working image tagged with the circle in the displayed stream of the video ?

PS: in my locally installed webots, the videos of the lower cameras are reversed on the screen :
On the left the high camera image is the participant's and the low camera is the opponent's and vice versa on the right !

[Jean-Eudes-le-retour]

what does the camera.send_to_robot_window() function do ?

The camera.send_to_robot_window(img) takes an OpenCV img argument and formats it to send it to a compatible robot window.

I forgot to add a working robot window to Eve's files, you can now pull the latest changes in Eve's repository and you should see a new window pop up when you open the world file. You can then go to the "Image feed" tab and when the controller will run you will see the image processing result on top of the robot's camera images.