mirror of
https://github.com/m1ngsama/robot_arm.git
synced 2026-03-26 04:03:49 +00:00
- README.md: comprehensive bilingual (CN+EN) rewrite Consolidates 使用说明书.md + 项目介绍文档.md into one document. Sections: overview, architecture, BOM, installation, quick start, voice commands, calibration, troubleshooting, technical notes, training reference, project structure. - TRAINING.md: rename from lora.md; add bilingual header. Full QLoRA fine-tuning research notes preserved as-is. - Delete: ck.md (dev journal), 使用说明书.md, 项目介绍文档.md All useful content merged into README.md. - .gitignore: remove stale whitelist entries for deleted files.
310 lines
12 KiB
Markdown
310 lines
12 KiB
Markdown
# 智能语音机械臂 / Voice-Controlled Robot Arm
|
||
|
||
基于"耳-脑-眼-手"全链路闭环的具身智能系统,运行于消费级硬件,完全离线。
|
||
|
||
*A full-stack embodied AI system — voice in, physical action out — running entirely offline on consumer hardware.*
|
||
|
||
---
|
||
|
||
## 系统简介 / Overview
|
||
|
||
| 能力 | 实现 | Capability |
|
||
|:---|:---|:---|
|
||
| **听** | Faster-Whisper,本地中文语音识别 | Speech-to-text (Chinese, local) |
|
||
| **想** | DeepSeek-R1-1.5B + QLoRA 微调,自然语言→JSON | LLM + rule engine, NL→JSON actions |
|
||
| **看** | YOLOv8s 目标检测 + 单应性矩阵手眼标定 | Object detection + hand-eye calibration |
|
||
| **动** | D-H 逆运动学 + S-Curve 轨迹规划,ESP32 驱动 | IK solver + smooth trajectory → ESP32 PWM |
|
||
|
||
硬件总成本 **¥317**,GPU 需求 RTX 3060 6GB(推理 <4GB 显存,延迟 <200ms)。
|
||
|
||
*Total hardware cost ¥317 (~$45 USD). Requires an NVIDIA GPU for LLM inference.*
|
||
|
||
---
|
||
|
||
## 系统架构 / Architecture
|
||
|
||
```
|
||
麦克风 / Microphone
|
||
│
|
||
▼
|
||
┌──────────────────┐
|
||
│ Faster-Whisper │ 语音识别 (STT) — 中文语音 → 文本
|
||
└────────┬─────────┘
|
||
│ "把削笔刀抬起5厘米"
|
||
▼
|
||
┌──────────────────┐
|
||
│ 规则解析引擎 │ 简单指令直接匹配(松开 / 复位 / 方向移动)
|
||
│ (Regex engine) │ 命中 → 直接生成 JSON,跳过 LLM
|
||
└────────┬─────────┘
|
||
│ 未命中(含物体名的复杂指令)
|
||
▼
|
||
┌──────────────────┐
|
||
│ DeepSeek-R1-1.5B │ QLoRA 微调推理
|
||
│ (QLoRA, FP16) │ 自然语言 → 结构化 JSON 指令
|
||
└────────┬─────────┘
|
||
│ [{"action": "lift", "target": "part", "height": 50}]
|
||
▼
|
||
┌──────────────────┐
|
||
│ YOLOv8s │ 实时检测目标物体
|
||
│ + Homography │ 像素坐标 → 机械臂工作坐标 (mm)
|
||
└────────┬─────────┘
|
||
│ (rx=170, ry=3)
|
||
▼
|
||
┌──────────────────┐
|
||
│ 运动控制引擎 │ D-H 逆运动学 + S-Curve 插值
|
||
│ arm_main.py │ 平滑轨迹 → 串口 → ESP32 → 舵机
|
||
└──────────────────┘
|
||
```
|
||
|
||
---
|
||
|
||
## 硬件清单 / Bill of Materials
|
||
|
||
总计 **¥317** / ~$45 USD
|
||
|
||
| # | 物品 | 规格 | 数量 | 单价 | 合计 |
|
||
|:--|:---|:---|:--:|---:|---:|
|
||
| 1 | 3D 打印机械臂(散件) | 教具级,含亚克力/PLA 结构件 | 1 | ¥71 | ¥71 |
|
||
| 2 | ESP32 开发板 | WiFi+蓝牙双核 MCU | 1 | ¥19 | ¥19 |
|
||
| 3 | ESP32 配件 | 接插件/扩展板 | 1 | ¥5 | ¥5 |
|
||
| 4 | USB 工业摄像头 | 免驱,广角,1280×720 | 1 | ¥61 | ¥61 |
|
||
| 5 | 数字舵机 MG996R | 金属齿轮,高扭矩 | 5 | ¥27 | ¥133 |
|
||
| 6 | 稳压电源 | 6V 6A,舵机专用 | 1 | ¥29 | ¥29 |
|
||
|
||
**硬件连接 / Wiring**
|
||
|
||
- **ESP32 串口引脚**:X→14, Y→4, Z→5, B→18, 夹爪→23
|
||
- **电源**:舵机与 ESP32 分开供电(外部 6V/6A),防浪涌
|
||
- **摄像头**:USB,固定于机械臂前方,覆盖整个工作台面
|
||
- **串口**:USB 连接 ESP32,默认 `COM3`,可通过环境变量 `ROBOT_PORT` 修改
|
||
|
||
---
|
||
|
||
## 安装 / Installation
|
||
|
||
### 1. 烧录固件 / Flash Firmware
|
||
|
||
Arduino IDE 2.x,开发板选 "ESP32 Dev Module":
|
||
|
||
```bash
|
||
# 打开 main.ino,选择正确串口,上传
|
||
# Open main.ino, select port, Upload
|
||
```
|
||
|
||
### 2. Python 环境 / Python Setup
|
||
|
||
Python 3.10+,CUDA 11.8 或 12.x(推荐)。
|
||
|
||
```bash
|
||
# 1. PyTorch(先去 pytorch.org 选对应 CUDA 版本)
|
||
# Visit pytorch.org to install the correct CUDA build first
|
||
|
||
# 2. 其余依赖 / Other dependencies
|
||
pip install -r requirements.txt
|
||
```
|
||
|
||
### 3. 配置 / Configure
|
||
|
||
所有可调参数集中在 `config.py`,支持环境变量覆盖:
|
||
|
||
```bash
|
||
# 修改串口(Windows COM 号 / Linux /dev/ttyUSB0)
|
||
# Change serial port
|
||
ROBOT_PORT=COM5 python voice_main.py
|
||
|
||
# 修改模型路径 / Change model paths
|
||
LLM_MODEL_PATH=D:\models\my_lora python voice_main.py
|
||
YOLO_MODEL_PATH=runs/best.pt python voice_main.py
|
||
```
|
||
|
||
默认值见 `config.py`,无需修改代码。
|
||
*Default values are in `config.py`; no code changes needed for standard tuning.*
|
||
|
||
### 4. 模型准备 / Models
|
||
|
||
**语音 (Whisper)**:无需准备,首次运行自动下载 `base` 模型。
|
||
*Auto-downloaded on first run.*
|
||
|
||
**视觉 (YOLO)**:需自行训练,50 张样本即可迁移学习:
|
||
|
||
```bash
|
||
# 用 LabelImg 或 Roboflow 标注你的物体,然后:
|
||
yolo detect train model=yolov8s.pt data=data.yaml epochs=100 imgsz=640
|
||
# 产出 runs/detect/train/weights/best.pt → 复制到项目根目录
|
||
# Copy runs/detect/train/weights/best.pt to project root
|
||
```
|
||
|
||
**大模型 (LLM)**:需要对 DeepSeek-R1-1.5B 或 Qwen1.5-1.8B 进行 LoRA 微调。
|
||
*Requires LoRA fine-tuning. See [`TRAINING.md`](TRAINING.md) for the complete guide.*
|
||
|
||
训练数据格式(Alpaca):
|
||
```json
|
||
{
|
||
"instruction": "把削笔刀抬起5厘米",
|
||
"input": "",
|
||
"system": "你是机械臂JSON转换器...",
|
||
"output": "[{\"action\": \"lift\", \"target\": \"part\", \"height\": 50}]"
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
## 快速上手 / Quick Start
|
||
|
||
```bash
|
||
python voice_main.py
|
||
```
|
||
|
||
启动后依次加载:机械臂串口 → YOLO 模型 → Whisper → LLM,弹出摄像头窗口。
|
||
*On startup: serial → YOLO → Whisper → LLM → camera window.*
|
||
|
||
**键盘快捷键 / Keyboard Shortcuts**
|
||
|
||
| 按键 | 功能 | Function |
|
||
|:---|:---|:---|
|
||
| **SPACE(按住)** | 录音,松开即识别 | Hold to record, release to recognize |
|
||
| **C** | 进入 / 退出手眼标定模式 | Toggle hand-eye calibration mode |
|
||
| **R** | 手动复位到原始姿态 | Manual reset to home position |
|
||
| **O** | 强制张开夹爪 | Force open gripper |
|
||
| **Q** | 退出程序 | Quit |
|
||
|
||
---
|
||
|
||
## 语音指令 / Voice Commands
|
||
|
||
所有指令用普通中文说话即可,无需特殊格式。
|
||
*Speak natural Chinese. No special syntax required.*
|
||
|
||
**抓取与搬运(需视觉定位)**
|
||
```
|
||
"把削笔刀抓起来"
|
||
"抓住那个盒子"
|
||
"把削笔刀抬起5厘米"
|
||
"将零件举高10公分"
|
||
```
|
||
|
||
**空间运动控制(精确移动)**
|
||
```
|
||
"向上三厘米" → Z 轴 +30mm
|
||
"向左移动四毫米" → Y 轴 +4mm
|
||
"往前伸10厘米" → X 轴 +100mm
|
||
```
|
||
|
||
**模糊移动**(不指定数值,默认 5cm)
|
||
```
|
||
"向左" "抬起" "往下"
|
||
```
|
||
|
||
**动作交互**
|
||
```
|
||
"点头" → 当前位置上下往复 3 次(幅度 3cm)
|
||
"摇头" → 当前位置左右往复 3 次(幅度 3cm)
|
||
"放下" → 降至桌面高度(Z=-15mm)并松开夹爪
|
||
"复位" → 回到初始安全姿态 [120, 0, 60] mm
|
||
"松开" → 张开夹爪,不移动
|
||
```
|
||
|
||
**语音兼容性**
|
||
系统内置谐音纠错:`"零米"→"厘米"`, `"小笔刀"→"削笔刀"`, `"电头"→"点头"` 等。
|
||
*Built-in homophone correction for common Whisper mishearings.*
|
||
|
||
---
|
||
|
||
## 手眼标定 / Hand-Eye Calibration
|
||
|
||
摄像头移动后必须重新标定。按 **C** 键进入标定模式:
|
||
|
||
```
|
||
依次点击 4 个角点 / Click 4 corner points in order:
|
||
|
||
P1 (左上) ←→ 机械臂坐标 (90, 90)
|
||
P2 (右上) ←→ 机械臂坐标 (200, 90)
|
||
P3 (右下) ←→ 机械臂坐标 (200, -90)
|
||
P4 (左下) ←→ 机械臂坐标 (90, -90)
|
||
```
|
||
|
||
点完第 4 个点后,单应性矩阵立即更新,无需重启。
|
||
*Homography matrix updates instantly after the 4th click. No restart needed.*
|
||
|
||
---
|
||
|
||
## 故障排除 / Troubleshooting
|
||
|
||
| 现象 | 原因 | 解决 |
|
||
|:---|:---|:---|
|
||
| 按空格无反应 | 窗口焦点不在摄像头画面 | 点击一下摄像头窗口 |
|
||
| 语音识别乱码 | 麦克风噪声 / 语速过快 | 安静环境,语速适中,按住空格 0.5s 再说话 |
|
||
| "未找到目标" | YOLO 未检测到物体 | 调整物体角度、光照;检查物体是否在训练类别中 |
|
||
| 抓取位置偏离 | 摄像头被移动 | 按 **C** 重新四点标定 |
|
||
| 无法连接串口 | ESP32 未插入 / 端口号不对 | 检查设备管理器,修改 `ROBOT_PORT` 环境变量 |
|
||
| 机械臂启动剧烈抖动 | 五路舵机同时上电浪涌 | 已在固件中处理(阶梯式上电),若仍出现检查电源容量 |
|
||
|
||
---
|
||
|
||
## 核心技术要点 / Technical Notes
|
||
|
||
以下是开发过程中解决的关键工程问题,供复刻者参考。
|
||
|
||
**D-H 逆运动学**
|
||
长度 130mm 的 L4 连杆导致几何解析法在水平移动时产生 40° 轨迹偏移。最终采用 Scipy SLSQP 数值优化器,加入 `Pitch=-90°` 姿态约束(抓手始终垂直地面),彻底解决非线性偏移。
|
||
|
||
*The 130mm L4 link caused ~40° path deviation with geometric IK. Solved by Scipy SLSQP numerical optimization with a Pitch=-90° constraint (end-effector always perpendicular to table).*
|
||
|
||
**S-Curve + 多层减震**
|
||
MG996R 在长力臂下惯性震动严重。减震流水线:倾斜补偿 → 移动平均滤波(deque)→ 速度限制 → EMA 阻尼 → 死区过滤。
|
||
|
||
*MG996R servos vibrate badly with a long lever arm. Solution: 5-layer damping pipeline — tilt correction → moving average (deque) → speed cap → EMA damping → dead-zone filter.*
|
||
|
||
**双通道解析架构**
|
||
简单指令(松开、复位、方向移动)走正则规则引擎,微秒级响应,且避免大模型将"向下三厘米"误判为 `lift`。只有含物体名的复杂指令才交给 LLM(延迟 <200ms)。
|
||
|
||
*Simple commands (release/reset/directional) bypass the LLM entirely via a regex engine (microseconds). Complex commands with object names go to the LLM (<200ms). This prevents the common failure mode of "move down 3cm" being misclassified as a lift action.*
|
||
|
||
**Pre-filling 截断**
|
||
DeepSeek-R1 的推理模型默认会输出思维链(`<think>...</think>`)。通过手动追加 `<|Assistant|>` 标签进行 Pre-filling,强制模型跳过思考过程直接输出 JSON,实现 100% 格式遵循率。
|
||
|
||
*DeepSeek-R1 defaults to outputting a chain-of-thought. Pre-filling with `<|Assistant|>` forces the model to skip the thinking phase and output JSON directly, achieving 100% format compliance.*
|
||
|
||
**Whisper 反幻觉**
|
||
三道防线:① 音频首尾静音裁剪;② `condition_on_previous_text=False`;③ 重复模式正则检测(去除"向右向右向右..."类幻觉)。
|
||
|
||
**工程坑:System Prompt 对齐**
|
||
训练与推理的 System Prompt 必须完全一致,否则模型输出偏移(如输出 500mm 而非 50mm)。已在代码注释中标注警告。
|
||
|
||
---
|
||
|
||
## 大模型训练 / LLM Training
|
||
|
||
约 500 条领域数据,QLoRA 微调 DeepSeek-R1-1.5B,Loss 收敛至 0.0519,格式错误率 0%。
|
||
|
||
完整训练流程见 [`TRAINING.md`](TRAINING.md),包括:QLoRA 超参数配置、GGUF vs Transformers 方案对比、Pre-filling 推理方案详解、实验结果。
|
||
|
||
---
|
||
|
||
## 项目结构 / Project Structure
|
||
|
||
```
|
||
robot_arm/
|
||
├── README.md 本文档 / This file
|
||
├── TRAINING.md 大模型 LoRA 微调研究笔记 / LLM fine-tuning notes
|
||
├── requirements.txt Python 依赖 / Dependencies
|
||
├── config.py 硬件与运动参数(支持环境变量覆盖)/ Hardware & motion constants
|
||
│
|
||
├── main.ino ESP32 固件,LEDC PWM 舵机控制 / ESP32 firmware
|
||
├── arm_main.py 机械臂运动学核心:D-H IK + S-Curve / Kinematics & control
|
||
├── whisper_main.py 语音识别封装 / ASR wrapper
|
||
└── voice_main.py 主程序:语音→LLM→视觉→控制 / Main app orchestrator
|
||
```
|
||
|
||
---
|
||
|
||
## 关键数据 / Key Specs
|
||
|
||
| 指标 | 值 |
|
||
|:---|:---|
|
||
| 硬件成本 | ¥317 |
|
||
| GPU 需求 | RTX 3060 6GB(推理 <4GB 显存) |
|
||
| 推理延迟 | <200ms(LLM),<50ms(规则引擎) |
|
||
| 训练数据量 | ~500 条 |
|
||
| 格式错误率 | 0% |
|
||
| 运行模式 | 完全离线 / Fully offline |
|