“Section 4.1 Terms and Definitions" has been deleted

Hello RCJ Rescue Committee,

Thank you for sharing the 2026 rules.
I have read RCJRescueLine2026-final.pdf and RCJRescueMaze2026-final.pdf.
and I have some questions about it.

Section 4.1, which was included in the 2025 rules, has been removed. In particular, Section 4.1.5 described parts and modules that were not allowed to be used.

Does the removal of Section 4.1 mean that there are no longer any prohibited parts or modules (i.e., that any parts or modules may now be used)?

Are items such as line-following sensors, AI cameras, and OCR libraries, which were prohibited in Section 4.1.5 of the 2025 rules, now permitted for use in 2026?

I posted this question on the Forum on February 15, but I have not yet received a response.
I would greatly appreciate it if you could kindly provide an answer.

Best regards,
MASA

1 Like

Hello MASA,

Thank you for reading the draft rules carefully and for your questions.

Even though the old “prohibited parts/modules list” (2025 section 4.1.5) is no longer present, that does not mean “everything is allowed” in 2026.

In the 2026 Rescue Line rules, the restrictions are written in a more general way. In particular, the rules still prohibit using commercially produced robot kits or sensor components that are specifically designed or marketed to complete any single primary task of RoboCupJunior Rescue (i.e., “task-solver” modules). This is intended to keep the league educational and fair.

What this means for your examples

  • Line-following sensors
    • If you mean a ready-made “line follower module” sold/marketed for line following / RCJ Rescue, then it is still not permitted under the “task-solver component” restriction.
    • If you mean general-purpose reflectance/IR sensors (that students integrate and program themselves), those are typically considered normal sensors and are generally acceptable.
  • AI cameras
    • A general camera / vision sensor can be fine.
    • But an “AI camera” module that is marketed as a complete line-following solution (or a complete rescue-task solution) would fall into the same “task-solver component” category and would not be permitted.
  • OCR libraries
    • The 2026 rules allow the use of external code, as long as teams credit the original creators and the team can explain what they are doing (i.e., it is not just using a complete solution without understanding).
    • So an OCR library is not automatically prohibited in 2026 by “parts/modules” rules. The key expectations are proper credit and student understanding.
    • It is very important that students understand it very well and are able to explain it.

Important practical note

Because these cases depend heavily on what exactly the item is and how it is marketed, the committee usually answers with this principle:

Generic sensors and tools are fine; “single-task RCJ Rescue solver” modules are not.

Best regards,
Mahmoud Madi
2026 RCJ Rescue Committee

Hello @Mahmoud
Thank you for your reply.

You wrote the following:

Generic sensors and tools are fine; “single-task RCJ Rescue solver” modules are not.

I believe this corresponds to what is written in Rule 4.3.2 of the Rescue Line rules and Rule 4.3.5 of the Rescue Maze rules.

However, it is difficult for us to determine what should be considered “single-task RCJ Rescue solver” modules. We would appreciate it if you could provide some examples.

In the 2025 rules, Section 4.1.4 listed examples of modules that were allowed, and Section 4.1.5 listed examples of modules that were not allowed. This made the interpretation easier. However, these examples were removed in the 2026 rules, which makes it difficult for us to understand the boundary.

For example:

a. Using an AI camera (HuskyLens) only as a simple camera without using its AI functions.
b. Using the AI functions of an AI camera (HuskyLens).
c. Using a PixyCam only as a simple camera without using its special functions.
d. Using the line-following function of PixyCam.
e. Using the object recognition function of PixyCam.
f. Using TensorFlow with a dataset that was created and trained by someone else.
g. Using TensorFlow with a dataset that was created by someone else but trained by ourselves.
h. Using EasyOCR.

Would these be considered “single-task RCJ Rescue solver” modules?

Best regards,
MASA

Hello @Mahmoud

It has been almost a month and a half since my previous inquiry, but I have not yet received a response. I would appreciate it if you could provide an answer.

Best regards,
MASA

1 Like

Dear Masa,

To sum it up, any camera that has the functionality to solve part of the challenge (e.g., line following, etc… ) is not allowed.

And for any use of libraries used, it should be fine as long as you understand it well and are able to explain it and credit the author.

Regards,
Mahmoud

Hello @Mahmoud
Thank you for your response. I understand that your answer was intended as a reply to my previous question.

However, it is difficult for us to determine what should be considered “single-task RCJ Rescue solver” modules. We would appreciate it if you could provide some examples.

According to your answer, cameras with specialized functions are prohibited, so devices such as HuskyLens, PixyCam, and UNIT-V would not be allowed.
If so, then:

a. NG
b. NG
c. NG
d. NG
e. NG

On the other hand, libraries may be used as long as the team can explain what they used (even if they did not design or train them themselves),
so:
f. OK
g. OK
h. OK

Is this understanding correct?

Best regards,
MASA

I think that a big part of the issue is that the above statement is rather vague. Consider the process of line following. This would involves…

  1. Sensing the line
  2. Extracting features from the sensor data
  3. Determining corrective actions
  4. Actuating motors

As far as I know, no camera/sensor exists that does all of the above out of the box, but all cameras/sensors do at least part of the above. Where exactly do you draw the line?

An Openmv cam can do 1 out of the box, and can be coded to do 2, 3, and 4. Feature extraction (eg. blob detection, line transform) are available through libraries, and while it is possible for students to know how the algorithms in the libraries work (…it’s opensource), few actually do. Do we require the students to understand and be able to explain how the libraries work?

A Huskylens can do 1 and 2 out of the box. For feature extraction, the code is not opensource, and it’s not possible for a student to know how they work (…but they can probably make a good guess). Since they can’t explain how it works, does it mean that’s it’s prohibited?

IR sensor arrays are often marketed as a “line following module”, though they really are just a bunch of IR sensors mounted on a PCB. They only do 1 (sensing). Will they be prohibited based solely on their marketing as a “line following module”?

I think we need to…

  1. Look past marketing, and at what the sensor actually does when determining whether something is allowed or not.

  2. Set clear lines on what constitute a “task-solver”, and apply it equally to both hardware and software. If using the line detection function in huskylens is not allowed on the basis of the students not understanding how it works, then students should also be prohibited from using the line transform function in libraries like openmv / opencv unless they can explain the algorithm.

1 Like

Generic sensors and tools are fine; “single-task RCJ Rescue solver” modules are not.

4.2.3. Teams are not permitted to use commercially produced robot kits or sensor components specifically designed or marketed to complete any single primary task of RoboCupJunior Rescue.

I think this rule is very ambiguous. I would appreciate it if the wording of this rule could be clarified, and if examples of permitted solutions and prohibited solutions could be listed to make it easier to understand.

Best regards,
MASA