[Betatesting] Follow road exercise

[javizqh]

Share here any issues, not solutions, related to this exercises. Do not comment anything else other than that.

This issue is not meant to be assigned to a single person, as its purpose is that multiple people test it.

[JitinSaxenaa]

Hi @javizqh @jmplaza

While testing the Drone Follow Road exercise, I noticed a couple of usability issues that might affect first-time users:

1 - Initial camera view – The starting camera angle is quite far from the drone, which makes it difficult for new users to quickly locate the drone’s position when the exercise begins.
2 - Restart behavior – Similar to other drone exercises, when restarting the exercise the drone does not pause and reset cleanly to the initial starting position. Because of this, we often need to restart the exercise multiple times just to locate the drone again and properly continue testing the code. [ Can i help in resolving this issue? As the changes might need to be done in RI repo , drone files. ]
3 - Gazebo simulation occasionally freezes – Sometimes the Gazebo simulation appears to get stuck: the camera angle or scene does not update on moving the cursor , After a short while it continues normally.

Demo Video

[kavyagupta100605]

Hi @javizqh,

I tested the Drone Follow Road exercise locally using RoboticsAcademy.

Environment

• OS: Windows
• Setup: Docker Desktop with RoboticsAcademy running locally
• Compose file used: user_humble_cpu.yaml

Steps performed

1. Started the RoboticsAcademy environment using Docker.
2. Opened the RoboticsAcademy interface in the browser.
3. Selected the Drone Follow Road exercise.
4. Clicked the Play button to start the simulation.
5. Waited for the simulation to initialize and checked the camera panels.
6. Tested the Restart functionality from the toolbar.

Observations

• The Gazebo simulation loads successfully and the environment appears in the Gazebo window.
• The exercise status changes to application_running, indicating that the backend connects correctly.
• However, both camera panels display "No image available" even after clicking Play and waiting for the simulation to initialize.
• I also tested the Restart button: the Gazebo simulation resets correctly, but the camera panels remain unavailable after restarting.

Browser console
I checked the browser developer console and noticed the following message when the exercise starts:

Failed to load resource: the server responded with a status of 404 (Not Found)

Other console messages indicate that the Dockerized application starts successfully and the exercise status becomes application_running.

Expected behavior
The camera panels should display the drone camera or road view when the exercise is running.

Screenshot attached below for reference.

Please let me know if additional logs or further testing would be helpful. I would also be happy to test other exercises if that would assist with the betatesting process.

[javizqh]

It is normal that it is shown that there are no images because you have not written any code to send the images.

[kavyagupta100605]

Thanks for the clarification!

That makes sense — I understand now that the camera panels remain empty until code is written to send the images.

I’ll continue exploring the exercise and testing the environment to better understand how the camera data is handled.

Please let me know if there are other aspects of the exercise that would be useful to test.

[tony19053000]

Hello @javizqh,

My name is Aayush Kumar, and I tested the Drone Follow Road exercise.

I was able to get it working correctly in the exercise. The ventral camera is detecting the road properly, the mask and center/error calculation are working, and the drone is able to follow the road using that control logic. I tested the full flow from my side and it is working properly.

I am also attaching a small demo video of my current working solution here.

Demo Video

[javizqh]

All of the reset and drift issues have been solved in JdeRobot/RoboticsInfrastructure#675 and will be available in the next Robotics Academy version (v6.1.3)

[evadrake89]

Hello Maintainers @javizqh @jmplaza ,

I have started working on the Follow Road exercise, and I feel that for someone who is a complete beginner, the user guide is a bit difficult to understand. The same issue occurred with the Line Follow exercise as well.

In my opinion, the following improvements could be made to the user guide:

Providing starter code for Python would be helpful. Simply explaining the lines is not sufficient, we need something like the example below:

import HAL
import Frequency
import WebGUI

HAL.takeoff(2.0)  # This will take the drone to a vertical height of 2 m

while True:  # A loop for continuously refreshing the image
    WebGUI.showImage(HAL.get_frontal_image())  # Front camera
    WebGUI.showLeftImage(HAL.get_ventral_image())  # Downward camera

Adding to this I am still not able to understand the importance of the Frequency module and the significance and use of this line:

Frequency.tick(ideal_rate)

[evadrake89]

Hello Maintainers @javizqh @jmplaza ,

If I get this in Gazebo simulation, how to restart the simulation from this point.....

I mean I am asking this because I have tried every single thing, but literally I am not able to restart this simulation.....

[javizqh]

Click on the Stop Universe button and select the Universe again, but that will probably not fix your problem.

Which are your computer specs and what steps did you do to get to that state?