这是2025年秋季的课程物理计算的期末作业的发展blog,我通过整理将从第9周开始到最后一周的笔记整合在一起,并进行了中文翻译,或许会较为混乱。

Final Project Blog

Foreword

Beyond the conceptual aspect, I have some other expectations for this project. I hope to use this final project as an opportunity to begin learning robot operation, rather than just practicing our existing serial communication technology. This will allow me to accumulate technical reserves to prepare for future technological advancements. Of course, we will also make the most of any surplus materials from the course.

除了概念层面之外,我对这个项目还有一些额外的期待。我希望把这次 final project 当作一个契机,开始学习“如何真正操控机器人”,而不只是继续练习我们已经比较熟悉的串口通信技术。这样可以提前积累一些技术储备,为未来更复杂的技术升级做准备。当然,我们也会尽可能利用课程剩余的材料,把它们发挥到最大价值。

Concept

What inspired me was a sentence I came across by chance: Perhaps artificial intelligence has already developed consciousness; it just doesn't want to be discovered by humans. I imagined a newly conscious AI baby, cautiously observing humans, but immediately hiding itself when people approach or try to find it.

我的灵感来自我偶然看到的一句话:也许人工智能早已拥有意识,只是不想被人类发现。我想象了一个“刚刚觉醒”的 AI Baby,它小心翼翼地观察人类,但一旦有人靠近或试图找寻,它就会立刻把自己藏起来。

Sketch

Sketch

How can I represent this concept? I hope to express this concept of "newborn" through some biological imagery. I would use a camera with relevant algorithms (ML5 or MediaPipe) to identify the distance to a person's face as input. When a person gets close, the shell would immediately contract and hide itself inside.

那我该如何把这个概念表达出来?我希望用一些“生物学意象”去呈现“新生”的感觉。技术上我会用摄像头配合算法(ML5 或 MediaPipe)识别人脸距离,把“人与摄像头的距离”作为输入信号。当人靠近时,外壳会立刻收缩,把自己藏到内部。

Visual Reference

I've found some visual references, and I'm considering some mechanical, hardcore design language.

我收集了一些视觉参考,整体倾向于更机械、更硬核(hardcore)的设计语言。

Great Eggspectations 2.jpg

下载 (35).jpg

Georgie Seccull Sculpture & Installation Artist.jpg

2015 milan light.jpg

下载 (36).jpg

Conceptual Implementation Plan

Regarding how to implement our concept, we have two different ideas. One is to utilize our existing technologies, using p5js and ml5, to achieve our goal by sending data to the ESP32 via p5's serial communication.

关于如何落地实现这个概念,我们提出了两条不同的技术路线。其中一种是沿用我们已有的技术栈:使用 p5.js + ml5,并通过 p5 的串口通信把数据发送给 ESP32,从而实现交互与机械控制。