Week 7 Overview

During this seventh week, I continued investigating the issue with the platform’s engine, RADI 4, within the Unibotics platform.

After studying the general structure and functionality of the engine’s Docker container, I began to review the engine’s execution in more detail. By accessing the Docker container internally, I could use the container’s native terminal.

Within the terminal, I manually executed the Gazebo program used for exercise simulation. In some attempts, I received no feedback from the terminal, and the program simply didn’t run. However, in other attempts, I encountered two errors:

libegl warning: egl: failed to create dri2 screen
228: invalid egl device

Upon investigating these errors, I realized they were related to the GPU exposure to the Docker container. The issue was that the Nvidia GPU was not properly exposed to achieve the desired graphical acceleration. Consequently, the Docker container couldn’t communicate effectively with the GPU, leading to the improper execution of the Gazebo program.

I have communicated these errors to the development team to make them aware of the issue.