Making industrial robots usable by the people on the floor 让车间里的人,也用得动工业机器人
My last role was deep inside a leading industrial AMR (autonomous mobile robot) manufacturer, building the software layer between powerful machines and the operators who run them — fleet scheduling, robot integration, and operator-facing interfaces. 我上一段深扎在一家领先的工业 AMR(自主移动机器人)厂商,做的是"强大的机器"和"操作它的人"之间那层软件——车队调度、机器人对接、面向操作员的界面。
The problem问题
An industrial robot — an AGV fleet, a robotic arm — is powerful, but its native interface speaks the language of engineers: coordinates, states, protocols. The person who actually runs it all day is a floor operator, not a roboticist. When the interface is intimidating or the integration is fragile, an expensive machine sits underused. 一台工业机器人——一支 AGV 车队、一条机械臂——很强大,但它原生的界面说的是工程师的语言:坐标、状态、协议。而真正整天操作它的,是车间工人,不是机器人专家。界面一旦让人望而生畏、或对接一旦脆弱,再贵的机器也只能低效闲置。
The constraint: this is where software meets a real machine约束:这是软件与真实机器的交界处
Software that commands physical motion is a different discipline from a web app. A dropped connection, an ambiguous state, or a mis-tapped control isn't a stack trace — it's a machine doing something in the physical world. It has to be safe, legible, and reliable on a real production floor. 驱动物理运动的软件,和写一个网页应用是两个学科。掉线、状态歧义、误触一个控件,换来的不是一行报错栈——而是一台机器在物理世界里动了起来。它必须在真实产线上做到安全、清晰、可靠。
What I built there我在那里做的
- Fleet scheduling (FMS) — coordinating the tasks and routes of multiple robots so they cooperate instead of colliding or deadlocking. 车队调度(FMS)——协调多台机器人的任务与路线,让它们协同而不是相撞或死锁。
- Robot integration — connecting the control layer to higher-level systems over industrial protocols (Modbus/TCP and similar), reliably, on the shop floor. 机器人对接——通过工业协议(Modbus/TCP 等)把控制层可靠地接到上层系统,在车间环境里稳定运行。
- Operator interfaces (HMI) — screens that let a non-technical operator monitor and control robots safely, with the safety-critical logic kept on the controller side and the UI as a clean, decoupled layer. 操作员界面(HMI)——让非技术操作员安全地监控和控制机器人的界面,安全关键逻辑留在控制器侧,UI 作为一层干净、解耦的表层。
- Industrial-grade rigor — high test coverage on distributed Go/Rust components, careful state handling, and disaster recovery (I once recovered 100% of lost production data by reverse-engineering MySQL Binlogs during an emergency). 工业级严谨——分布式 Go/Rust 组件的高测试覆盖、审慎的状态处理、以及灾备(我曾在应急中逆向 MySQL Binlog、100% 恢复丢失的生产数据)。
Why this is my kind of problem为什么这正是我擅长的问题
It sits exactly on the seam between software and hardware — where a clean interface has to sit on top of a physical, unforgiving machine, and where you earn the operator's trust by respecting what the hardware can do. It's the same instinct behind everything I build: take something only an expert can operate, and make it safe and obvious for the person who actually has to use it. 它恰好落在软件与硬件的交界线上——一个清爽的界面之下,是一台物理的、不容出错的机器;你靠尊重硬件的能力边界,赢得操作员的信任。这和我做的每件事背后是同一种直觉:把一件只有专家才玩得转的东西,变得让真正要用它的人既安全、又一目了然。
Proof证明
Shipped in production at a leading industrial AMR manufacturer. My engineering standards and integrity there are backed by a formal recommendation letter from the company's CTO — happy to share it on request. 在一家领先的工业 AMR 厂商交付于生产环境。我在那里的工程标准与职业操守,有该公司 CTO 的正式推荐信背书——可应请求分享。
Where else this applies还能用在哪
Any operation where an expensive machine is underused because only an engineer can drive it: factory robots and arms, industrial equipment, IoT / PLC systems, custom automation. If your hardware is powerful but hard for your team to actually operate, that gap is exactly what I close. 任何"因为只有工程师会开、导致贵机器被闲置"的场景:工厂机器人与机械臂、工业设备、IoT / PLC 系统、定制自动化。如果你的硬件很强,却难被团队真正用起来——我补的正是这道缝。
← All case studies← 全部案例