Introduction & Guidance on Next Steps

[keerthivnair]

Hi everyone,

I’ve been exploring JdeRobot’s exercises and have worked through the Follow Line, Vacuum Cleaner and Digit Classification ones. Going through these gave me a good sense of how the exercises work at the surface level, but now I’d like to get a clearer understanding of the internal flow and structure.

From what I’ve seen, there are several moving parts — the HAL interfaces, the GUI/simulator side, the ML models, and the Dockerized environment that ties everything together. I’m trying to build a mental map of how these pieces interact:

where HAL interfaces are defined and connected,

how and where models are actually loaded/used inside the flow,

how the GUI or simulator communicates with the backend,

and finally, how ROS concepts (fit into this architecture ( I'm totally new to ROS would love to learn more and work with it).

I’ve skimmed through parts of the repositories and Docker setup, but I’d love some guidance on whether I’m looking in the right places and what’s the best way to deepen this understanding. If you have suggestions for specific modules, files, or even beginner-friendly issues to dive into, I’d be glad to explore them.

Thanks a lot — really excited to keep learning and contributing more meaningfully!

[jmplaza]

Hi @keerthivnair ,

nice to meet you and welcome to RoboticsAcademy community.

1.- the hal interface for each exercise is a Python module that the RoboticsAcademy user may use to access robot sensors and robot actuators (instead of using the original ROS topics). It is defined at each exercise directory. Currently we are moving RoboticsAcademy to support direct ROS2 topics programming, with the user directly managing the sensor and actuator ROS2 topics on each exercise if he/she is inclined to do so (@AshishRamesh may provide more info if needed).

2a.- the robot models and world models are defined in the RoboticsInfrastructure repo. They are both automatically loaded when entering on each exercise webpage.
2b.- The neural model for the DeepLearning exercises is uploaded by the user as part of his/her solution (@codezerro may further clarify this). The user has to explicitly upload them before running its Python source code for the exercise.

3.- each exercise has its own webgui.py Python module, which provides functions that can be used from the user code to show data at the web browser (for instance a processed image in the FollowLIne exercise) and to read data from the web page (for instance the mouse click of the user inside the city map in the Global Navigation exercise). Just check user documentation of those exercises.

4.- the simulator is run inside the RADI (=RoboticsAcademy Docker Image) which contains all the robotics dependencies preinstalled, and communicates with the web browser through a VNC connection. The RADI has to be downloaded and launched before start using RoboticsAcademy.

5.- regarding ROS, I suggest you to read the online documentation for ROS2 available at many ROS2 websites.

Beyond user documentation, you have also developer documentation. Probably they can be improved, feel free to further ask here any question :-)