Regarding the Use of Products with AI Capabilities

RoboCupJunior Rescue RoboCupJunior Rescue Maze
1

Hello,
According to the section “Terms and Definitions of the Robot,” particularly clause 4.1.3, products such as AI cameras are listed as prohibited examples if they are capable of acquiring information without further development.
However, the term “AI Camera” is widely used in camera vendors selling/promoting their products. This raises a concern: if a product has “AI” in its name, is it automatically prohibited, even if we don’t use non-coding informatioin acquisition or camera itself doesn’t have such capabilities to build model without coding?
I would appreciate it if you could clarify the specific criteria that determine whether a product is prohibited.

Currently, we are using the UnitV K210 “AI Camera” from M5Stack for victim recognition. We are using a custom-trained model and have programmed model generation by myself. We are making micro python code running on this device using OpenMV-based API for inference/prediction. So we had enough development and not used any non-coding information acquisition/model generation.
In this use case, would the use of this camera be allowed?

3 Likes
2

We are also using the same camera (UnitV AI camera), so we are very interested in this discussion!
However, it has been a month since the post, and there is still no response, which makes us a bit anxious.
As a reference, we will describe how our team is using this camera.
We identify the presence of color and letter victims by applying thresholds (color, area) to the captured image. If a letter victim is detected, the image is fed into a CNN model to classify which victim it is (H, S, U, dummy).
This CNN model is based on a MobileNet architecture, which has been modified to suit this competition and trained using machine learning.

Best regards,
Haruki Tokumitsu

2 Likes