Embedded Systems · Robotics · AI-native Engineering
嵌入式系统 · 机器人 · AI 原生工程
Automation undergraduate at SWUST (Class of 2027)
西南科技大学自动化专业本科生（2027 届）
RT-Thread · Motion Control · Computer Vision · Planning & Solvers · Agentic Systems
I build systems that connect algorithms to physical machines: embedded firmware, motion control, machine vision, planning, communication protocols, and the tools needed to debug them as one system.
我专注于把算法真正落到物理设备上：从嵌入式固件、运动控制、机器视觉和任务规划，到通信协议与整机联调工具，尽量把完整链路做通、做稳、做成可复现的工程。
I served as team lead for two seasons of the National University Student Intelligent Car Competition. In 2026, our team won the National First Prize in the Smart Vision category.
我连续两届担任全国大学生智能汽车竞赛视觉组队长。2026 年，西科视觉九队获第二十一届全国大学生智能汽车竞赛智能视觉组 全国一等奖。
| Year | Competition | Result |
|---|---|---|
| 2026 | 21st National University Student Intelligent Car Competition, Smart Vision / 第二十一届全国大学生智能汽车竞赛智能视觉组 | National First Prize / 全国一等奖 · Team Lead / 队长 |
| 2026 | NUEDC Regional Contest, Problem E: Puzzle Device / 全国大学生电子设计竞赛赛区赛 E 题“拼图装置” | Sichuan First Prize / 四川省一等奖（全校唯一） |
| 2025 | National Undergraduate Electronics Design Contest, Problem C: Monocular Vision Measurement / 全国大学生电子设计竞赛 C 题“单目视觉测量” | National Second Prize / 全国二等奖 |
| Project | What it does / 项目简介 | Core stack / 核心技术 |
|---|---|---|
| rt1064-sokoban-car | National First Prize smart car stack: a physical robot plays a vision-based Sokoban game with Classic, Semantic, and Bomb tasks. 全国一等奖智能视觉小车完整开源工程：实体小车通过视觉定位完成 Classic、Semantic 与 Bomb 推箱子任务。 |
RT1064, RT-Thread, C/C++, OpenART, AprilTag, motion control, task planning |
| E-puzzle-device | End-to-end puzzle device from fisheye vision and geometric solving to collision-aware gantry pick-and-place. 从鱼眼视觉、碎片识别和几何求解，到龙门架防碰撞取放的完整拼图装置。 |
MaixCAM, STM32F407, Python, C, OpenCV, path planning |
| phone-debugger | Android control and telemetry terminal built for real-vehicle debugging: map and pose visualization, virtual display, remote control, and online PID delivery. 面向实车联调的 Android 终端：地图与位姿可视化、虚拟屏同步、遥控和在线参数下发。 |
Kotlin, Jetpack Compose, UDP, protocol design |
| QuantAgent | Agent-driven quantitative research platform with an event-driven engine, historical replay, WFA robustness analysis, and pgvector RAG. Agent 驱动的量化研究平台，包含事件驱动引擎、历史回放、WFA 鲁棒性评估与 pgvector RAG。 |
Next.js, FastAPI, Go, NATS, PostgreSQL, pgvector, ClickHouse |
- Embedded & Control / 嵌入式与控制: RTOS firmware, motor control, sensor fusion, real-time diagnostics, and hardware-software integration.
RTOS 固件、电机控制、传感器融合、实时诊断与软硬件联调。 - Vision & Planning / 视觉与规划: geometric vision, pose mapping, search and optimization, collision-aware motion, and solver design.
几何视觉、位姿映射、搜索优化、防碰撞运动与求解器设计。 - AI-native Engineering / AI 原生工程: using coding agents for requirement decomposition, implementation, review, and evidence-based debugging.
使用 Coding Agent 完成需求拆解、实现、审查与基于证据的故障定位。 - Developer Tools / 开发工具: building the missing tools when integration friction becomes the bottleneck.
当联调效率成为瓶颈时，直接补齐协议、可视化、自动化与诊断工具。
- Trace behavior to live code, runtime state, and hardware evidence. / 以当前代码、运行状态和硬件证据确认真实行为。
- Keep experiments reproducible with explicit configuration, logs, and versioned milestones. / 用明确配置、日志和版本里程碑保证实验可复现。
- Treat debugging tools as part of the product, not disposable scripts. / 把联调工具视为系统的一部分，而不是一次性脚本。
- Prefer a complete, testable chain over an isolated algorithm demo. / 比起孤立算法演示，更关注完整链路是否可执行、可验证。
Maintaining the post-competition open-source release of rt1064-sokoban-car and continuing to build embedded, robotics, and agent-assisted engineering projects.
目前正在维护 rt1064-sokoban-car 的赛后开源版本，并继续探索嵌入式、机器人与 Agent 辅助工程开发。
- GitHub: @onemott
- Email: 1532439062@qq.com