This week, I started my guided practices in robotics. As a first task, I was told to recreate the Monaco circuit with realistic heights.
I started the job in Blender using the GIS add-on, which provides realistic maps, so I had a good template to make the circuit line.
I also installed the robotics simulator Gazebo (Harmonic), which needed a Linux distribution to run. In this case, Ubuntu provides a Windows
image that allows working with their system inside Windows.
Below is a top-view capture of the plane that GIS can create with the circuit line I made.
Week 2: 10-14 February, 2025
This week, I finished the first version of the circuit, so I went into Gazebo to create a world that includes the track.
I encountered a few small problems when I first entered the simulator, as it is a completely new environment to get used to.
For example, I couldn't import any custom objects, so I tried with a very simple custom cylinder before following the same process with the circuit.
This helped me learn about the worlds and formats that Gazebo uses to recreate reality.
I also started the process of including the circuit in the database, but so far, I haven't been able to do it safely.
That will be my work for next week, along with making a new circuit that you will be able to see in the next entry.
As I will do every week, here is a new image of my progress, in this case, a Gazebo world with the Monaco track.
Week 3: 17-21 February, 2025
This week, I was out for three days due to a fever, so I could only work for two days.
Now, I can reveal the circuit I have been working on: Spa, the Belgian track where F1 cars, pilots, and teams deliver one of the best races each year.
I also started learning about the system that robots use to work in Gazebo, which involves using plugins.
This week, I have not been very productive, but I am surely making progress in my work.
I've included a Blender image of the circuit, which I managed to texture using Google Maps satellite view.
Week 4: 24-28 February, 2025
This week, I experienced a setback as I could not open a Gazebo world using ROS 2.
I asked a colleague for help, so we started working together to solve the issues.
This led to the realization that I needed to change the computer's operating system. Until now,
I have been working with the Windows Subsystem for Linux (WSL), but in the end, this caused trouble with some program installations.
So, it is better to install a dual-boot system (to avoid erasing the Windows system I normally use).
Hopefully, next week I will be able to start working on the new task I have been assigned: an Ackermann robot that follows the race line on the circuits.
Here you can see Gazebo Harmonic running on my brand-new Ubuntu operating system.
Week 4: 3-7 March, 2025
This week, I continued working on Gazebo and ROS2, trying to get Gazebo to open with ROS2.
As I changed my OS to Ubuntu, programs are now running on their native system instead of running on the Windows Linux Subsystem, which caused some problems.
Now I am currently opening Gazebo with ROS2 with no problem. This will allow me to see more details of the simulation once I open a world with a robot.
As an Ackermann F1 robot is quite hard to make, I was recommended to take a simpler robot from the roboticsInfrastructure repository.
In this case, I took the Turtlebot robot, which is made out of a quad body and two wheels.
I've included an image of the Turtlebot inside an empty world.
Week 6: 10-14 March, 2025
This week, I slightly changed the approach to the Ackermann robot problem.
As progress was too slow, I talked to Pedro, an experienced person within the robotics structure we are working on.
He explained to me that it was not necessary to use ROS2 to implement the Ackermann car. In fact, it is easier to avoid ROS2 until the robot is complete.
Pedro also gave me some references to strengthen my knowledge base, which led me to explore the structure of .sdf files
much more than I had done before, the plugin system of Gazebo, and some repositories with Ackermann models.
Previously, I showed Spa in Blender, but here you can see it inside Gazebo.
Week 7: 16-20 June, 2025
Yes, you're not wrong. The date is correct.
On the last day of the previous week I posted, I got injured in my arm while playing football with my friends.
I have spent several weeks resting and some others in rehabilitation.
This week, I had to get back to work by remembering all I had done before, which may not seem like much, but the processes
I made before were hard to do, so there were some steps I missed when trying to recreate things I did before.
I also had to reintroduce myself to Ubuntu, but that was easy as Ubuntu is a very user-friendly OS.
Week 8: 23-27 June, 2025
This week, I started working with an Ackermann model a colleague created while I was injured.
I have been constantly trying to make a robot spawn in an open world, but for now, it has been impossible.
It doesn't matter if I use ROS2 to launch the world or if I use Gazebo directly; the car does not spawn.
I have been told to make an exportable object of the circuit. By doing this, the visualizer of the roboticsAcademy web can show a preview of the map.
This seems easy, as the visualizer works with .glb files, which can be exported from Blender. However, the reality is that you need
to make a double export: first to .dae and then to .glb. So, I made the documentation to keep that registered in the GitHub repository.
Week 9: 30-4 July, 2025
This week, I was able to spawn the car in both the Monaco and Spa circuits.
Even if this seems like good news, the car was having some trouble trying to avoid its wheels from going into the asphalt in Monaco.
As I didn't know what was happening, I moved to Spa because the approach to the build solution is different, and the car might not introduce its wheels there.
And as I suspected, in Spa, the car doesn't go through the road, so I could test the car a few times there.
After that, I tried to separate the car spawn code from the Spa launcher to an independent file. That didn't work, and for reasons I still don't know,
I couldn't go back and spawn the car again in the correct pose. It appears initialized at zeros.
Last but not least this week, I talked to Nikhil, the creator of the car, to ask him why the car didn't show the textures, and he told me that in that particular car,
the textures were not migrated to Harmonic, but he was working on another car with textures.
Week 10: 7-11 July, 2025
This week, I primarily worked on car integration.
Before going into that, I made some documentation of the world creation process in Blender, so other collaborators can make complex and realistic custom worlds.
Now, talking about the car, I didn't make good progress. The car still goes through the asphalt in Monaco and cannot spawn in the correct pose in Spa.
All these weeks, I have been experiencing bugs and malfunctions with Gazebo in Ubuntu. Each time I try something new, I have to reboot the computer to make the changes visible.
Normally it is only necessary to recompile the project, but this doesn't work in my case, so I will try to fix this for the future because it is causing me significant loss of time.
Week 11: 14-18 July, 2025
This week, I have been heavily stuck.
As I did last week, I worked on spawning and despawning the car in Gazebo using ROS2.
Right now, the car spawns without problems in the world. It bridges between Gazebo and ROS2, and executes the logic I put into it via console. Up to this point there are no problems
at all, but when it's time to despawn, the car and the car's topics don't disappear, even if the console command's return is correct.
I spoke with some colleagues, and we reached the conclusion that this situation is exceptional and is a bug, so I moved to the next step, which is trying to make the car follow the line.
I didn't have time to start, so I hope to make some progress on this next week.
To include a photo here, I am showing the buildings I added to Monaco so it looks a bit better.
Week 12: 21-25 July, 2025
This week, I went back to working on the car.
Fortunately, this time I understood the car far better than last time.
It turns out that the code to make the car despawn needs a confirmation that it is actually a model that needs to be despawned. I was writing all
the parameters the car has, including that it is a model, but at the end of the command, I had to add a redundancy.
Then I continued by focusing on the 'follow line' exercise I have to migrate from Gazebo classic to Harmonic. The next step is to create a ROS2 launch
file that will automatically bridge into the car, but I only started it so I can't show it this week.
Since almost all the work this week is code-based, there won't be a photo this week.
Week 13: 28-31 July, 2025
This week, I tried to make the launch file bridge the car directly.
It is harder than it seems and the worst part is that I don't know the reason why it is not working.
Right now, the topics seem to be working, but the bridge doesn't start, so it is impossible to give the car instructions either
by console or by script.
Since I am starting to get stuck again, I asked for some other work to do while trying to unlock the key to this exercise, so I was told to start
looking into the autoparking exercise, in which a car has to look for a place to stop and park there.
Week 14: 1-5 September, 2025
This week, I read and understood the autopark exercise.
From last week (and a month apart) I inherited the malfunction problem of the car's bridge.
As this was the first week of coming back to work after the summer break, I had to readapt and started remembering how everything worked.
To solve the bridge problem, I asked Javier, my colleague, for help, and he told me my ROS2 version was not ideal to work with harmonic.
He told me to change it or to use BT-Studio, a test site that includes the correct version by default.
I obviously started the installation when Ubuntu collapsed due to a lack of space. Sadly, the only option to solve that is a new Ubuntu installation
with more space, so I did it, saving all the work done before the reinstallation.
At least I could get into BT-Studio, so I don't have to worry anymore about program versions.
Week 15: 8-12 September, 2025
This week, I worked closely with my colleague Javier to try to solve an unexpected problem I encountered while using BT Studio.
The issue was that BT Studio simply refused to synchronize the GitHub pool with Visual Studio Code, preventing me from keeping my local work
aligned with the remote repository. We went through different possible solutions, from revising credentials and permissions to reinstalling
extensions and checking the workspace configuration.
After many hours of trial, error, and adjustments, all our efforts unfortunately proved to be in vain. The platform continued to block the
synchronization process, and the problem remained unresolved.
I decided not to waste any more time and to redirect my focus toward the Autoparking exercise, an activity that I had
been planning to address later. This exercise involves creating three different parking scenarios (angled parking, parallel parking, and perpendicular parking)
to test autonomous vehicle behavior. For this, I began by carefully reviewing how the entire system was originally implemented in Gazebo Classic, taking notes
and analyzing its structure so I can prepare an effective migration to Gazebo Harmonic.
Week 16: 15-19 September, 2025
This week, I created new maps for a migration of Autoparking exercise.
The Autoparking exercise has three different modes: angled parking, parallel parking, and perpendicular parking.
Since I currently cannot test with the available development tools, I will start working directly on the database, using RoboticsAcademy as my
primary test site.
To do that I cloned the repo and followed the user guide instalation of RA.
I actually had no time to test if that worked out to change the excercises, but next week I will see.
