SWCS is a compact, offline-capable recycling station for small venues — cafés, office pantries, labs, school micro-recycling points — where there's no staff to watch the bin and no network to rely on. It closes the full loop from sensing → recognition → decision → physical sorting, not just classification on a screen.

It was my final-year project for ELEC S411F (Electronic & Computer Engineering Project, 2025/26), supervised by Dr. Tong Kwong Bun Bruce.

Demo

The problem

Recycling quality collapses when materials are contaminated at the source. Most academic work stops at lab accuracy, and most commercial "smart bins" are large, expensive, network-dependent installations for big facilities. There's a real gap for something small, cheap, robust and offline that can actually act on its decision — and safely refuse to act when it isn't sure.

Approach

A two-stage edge-vision pipeline plus a confidence-gated controller:

• Localize — YOLOv8 detects the item and crops the region of interest (ROI).
• Classify — MobileNetV3-Large labels the ROI as paper / plastic / metal / glass / cardboard. Chosen for edge latency (not just accuracy) and deployable via ONNX / TorchScript / TensorRT / CUDA.
• Decide — a Decision Engine fuses the model probabilities and applies a confidence gate: divert only if the class and its confidence clear a threshold; otherwise the item stays on the default bypass route. Low-confidence cases never trigger a wrong physical action.

System architecture

• Upper machine: NVIDIA Jetson Orin Nano runs camera capture, YOLO localization, ROI cropping, classification and the Decision Engine.
• Lower machines: two Arduinos — belt_arduino (IR trigger + conveyor) and servo_arduino (sorting flap) — for clean separation of control.
• Protocol: USB-serial JSON with an identity handshake between the Jetson and each Arduino.
• Closed loop (FSM): IR sensor detects an item → stop belt → capture frames → YOLO + MobileNetV3 + confidence gate → JSON command → servo flap rotates → belt resumes → cycle logged with image, prediction, confidence, action and timing.

Training used PyTorch + PyTorch Lightning (OmegaConf config, torchvision ImageFolder) on a self-collected dataset that ranged from clean backgrounds to cluttered, real-world, varied-lighting scenes to harden against contamination and occlusion.

Results

A labelled 120-cycle closed-loop experiment (60 metal + 60 non-metal), using metal routing as the physical test case:

• Offline classifier accuracy: 93.59% — metal 93.69% precision / 95.59% recall / 94.63% F1
• System routing accuracy (2-class): 90.8% — metal precision 94.5%, recall 86.7%
• False diversion: 5.0% · Missed diversion: 13.3%
• Mean end-to-end cycle: 3.29 s — of which recognition was only ~61 ms

What I learned

• The AI was never the bottleneck. Recognition took ~61 ms; the full cycle ~3.29 s, dominated by servo actuation (~2.7 s) and conveyor recovery. The hard engineering was mechanical timing, not inference.
• Closed-loop accuracy depends on the actuator, not just the model. A high-torque servo caused PWM jitter and unstable flap angles — a correct classification could still mis-sort. An independent 5 V supply with a common ground to the Arduino removed the jitter entirely.
• Confidence gating is a trade-off. It kept false diversion low (5%) but raised missed metal diversions (13.3%) — a knob to tune against labelled system-level data, not just offline validation.

Further work

Multi-path routing for all five materials (not only the metal test path), latency reduction in the mechanical chain, and long-duration reliability testing — camera drift, serial stability, belt misalignment, servo wear.

最近我一直在认真打磨一款自己的产品 —— Car Boy，一个写给爱车人的社区。

For a while now I've been quietly polishing a product of my own — Car Boy, a community built for people who love their cars.

这是什么 · What it is

它不是那种只能刷帖的传统车友论坛。我想做的，是一个更完整、更精致、也更贴近爱车人日常的平台 —— 把路线、圈子、动态、车库、活动、油价这些高频场景，尽量收进同一个产品里。

It isn't the usual car forum where all you can do is scroll a feed. What I'm after is something more complete, more polished, and closer to how car people actually live day to day — bringing routes, clubs, posts, your garage, events, and fuel prices together into a single product.

现在能做什么 · What works today

现在已经做出了相当完整的一版：

There's already a fairly complete first version up and running:

• 发现并收藏喜欢的路线 / Discover and save the routes you love
• 查看路线详情、地图轨迹与图片 / Open route details, map traces, and photos
• 加入圈子、发布动态 / Join clubs and share posts
• 管理自己的爱车与个人主页 / Manage your cars and your personal profile
• 持续打磨更适合大中华区的体验 / An experience tuned for users across Greater China

截图都是真的 · The screenshots are real

这次放出来的几张宣传图，用的都是产品里的真实页面，不是凭空画的概念图。

The promo shots I'm sharing are all real pages from inside the app — not mockups or concept art.

我始终相信，真正好的车主产品不该只是「能用」，而应该有一点质感，也有一点热爱。Car Boy 还在持续迭代，但它已经越来越接近我最初想做的样子了。

I've always believed a great product for car owners shouldn't merely "work" — it should carry a little craft, and a little love. Car Boy is still evolving, but it's getting closer and closer to what I set out to build.

想听听你的想法 · I'd love your input

如果你也是车迷，喜欢跑路线、拍车、记录用车生活，欢迎来看看；也很想听听你最希望这类产品补上什么功能。

If you're into cars too — chasing good roads, photographing them, keeping a record of life with your car — come take a look. I'd genuinely love to hear which features you'd most want to see next.

几句坦白话 · A few honest notes

产品还在很早期，有些地方要请你多包涵：

It's still very early days, so there are a few things I'd ask you to bear with:

• 现阶段对发帖、发布路线、创建圈子的数量有所限制。 / For now there are limits on how many posts, routes, and clubs you can create.
• 受资金所限，网站目前跑在我家里的服务器上，访问速度偶尔会不稳定。 / Funds are tight, so the site currently runs on a server at home — speeds can be uneven at times.
• 早期阶段暂时只支持邮箱注册。 / In this early phase, sign-up is email-only.

感谢你的理解与包涵 😭

Thanks for your patience and understanding 😭

—— Hiko