推荐 5 款免费开源,零基础 DIY 智能 AI 桌面机器人,适配DeepSeek,豆包,Qwen大模型Recommended 5 Free and Open-Source Zero-Basis DIY Intelligent AI Desktop Robots, Compatible with DeepSeek, Doubao, and Qwen Large Models


AI free token/AI免费积分

 

图片
EchoEar 喵伴 AI 机器人搭载的 ESP-Brookesia 框架实现全双工语音交互、多模态识别与智能体控制,构建更具沉浸感的人机交互体验。

EchoEar 套件以端到端开发为核心理念,技术亮点如下:

  • 全双工语音交互:用户可以与设备进行连续对话,而不仅仅是单次响应。

  • 多模态识别与智能体控制:能够识别用户的多种输入方式,并提供相应的智能响应。

  • 情绪识别与动态表情互动:能够识别用户的情绪变化,并通过动态表情进行互动,增强设备的人格化表现。

  • 长记忆能力:支持对用户对话内容的持续记录,提供更具陪伴感的个性化体验。

  • 离线语音唤醒与声源定位:结合电机控制,实现180°范围内的方向跟踪,使交互更自然。

  • 智能家居控制:支持MCP协议与Function Call能力,可用于远程控制家庭设备,作为智能家居系统的本地中枢。

乐鑫开源EchoEar 喵伴智能 AI 机器人,可接入豆包、DeepSeek、Qwen 3、文心一言

 

 

 坦克人 ESP-SparkBot  

 

 

图片

ESP-SparkBot 是⼀款基于 ESP32-S3开发板,集成语⾳交互、图像识别、遥控操作和多媒体功能于⼀体的智能设备。它不仅可以通过语⾳助⼿实现⼤模型对话( DeepSeek、OpenAI 、通义千问)、天⽓查询、⾳乐播放等互动,还使⽤⼩度⼿机 APP 完成蓝牙配⽹、⾳⾊切换、⾳乐播放及其他智能服务。

 

同时,ESP-SparkBot 内置加速度传感器,⽀持摇⾊⼦和 2048 游戏等娱乐交互功能。硬件⽅⾯,其磁吸式设计⽀持模块扩展,可轻松转换为遥控⼩⻋,实现摄像头实时传输和⼿机操控。此外,设备⽀持本地 AI 处理,可以运⾏⼈脸识别和动作检测功能。你还可以通过投屏模块实现⾼清视频播放和游戏运⾏,展⽰强⼤的性能和多功能性。

  • 天气时钟,通过IP地址显⽰当地⽇期、天⽓、温度

  • 大模型AI对话,通过接入 DeepSeek、OpenAI 、通义千问等大语言模型,实现对话聊天功能

  • 蓝牙音箱,支持语音控制小度,支持APP中进行音量调节、切换播报音色、还能绑定QQ音乐/喜马拉雅播放音乐、新闻、有声读物……

  • 语音/人脸识别&移动检测,可本地离线运⾏,包含猫脸识别、⾏⼈检测等

  • USB 投屏,可作为电脑副屏、扬声器、麦克⻛使⽤,还能体验电竞、3A⼤作

  • 智能小车,支持手机/语音控制、无线图传、拍照,切换灯光功能,开发了配套的控制界面,使用浏览器打开即可使用

  • 赛博木鱼,敲赛博⽊⻥,即可积累功德,通过⼴播多个设备同时敲⽊⻥,还能成倍积累功德,功德圆满后进入放松摸⻥状态

  • 休闲小游戏,赛博骰子、2048 游戏等

图片
仅358元,用ESP32 开发板 DIY一个AI大模型瓦力机器人,可接入DeepSeek、OpenAI

 

 

 小智 EDA-Robot 

 

 

图片
EDA-Robot 小智AI机器狗采用乐鑫 ESP32-S3R8N16 模组,结合 INMP441 + MAX98357 实现小智AI 大语言模型对话及 MCP 协议控制机器狗运动,无需通过手机App操控, 只需唤醒小智执行 “向前走”、“向后走” 等口语化指令即可。
  • 内置AI大模型:支持小智AI提供的DeepSeek、Qwen、豆包等大语言模型,能进行自然语言理解与处理。

  • MCP协议适配:支持MCP协议服务,LLM可通过MCP协议控制机器狗动作。

  • 自然语言控制:支持通过自然语言控制,无需设置词条,均为LLM推理完成。

  • 自定义唤醒词:支持自定义名称唤醒机器狗,定制专属宠物。

EDA-Robot 小智AI机器狗主控选用 ESP32S3系列模组,集成 WiFi 功能,提供丰富的 GPIO 接口。 板载 USB 转串口芯片,方便程序下载和调试。

音频采集电路使用 INMP441 数字麦克风模块进行音频采集,结构简单,可根据需求替换为更低成本型号。

音频输出部分采用 MAX98357 模块,支持I2S音频输入,驱动扬声器输出。设计简单,性能稳定。

显示屏部分采用 SSD1306/SSD1315 OLED 屏幕,分辨率128×64,尺寸0.96或1.3英寸,需注意I2C线序及上拉电阻配置。

电源供电电路采用两节串联 14500 电池,串联后满电电压 8.4V, 通过两路 LDO 分别降压至 5V 为舵机供电和 3.3V 为 MCU 供电。

舵机电路采用 SG90/MG90 舵机,通过 PWM 信号控制角度。 由于舵机内部自带 5V 转 3.3V 驱动电路,PWM 信号可直接由 MCU IO 控制,无需电平转换。

按键电路与下载接口设计简洁,均可直接连接 MCU IO。由于 MCU 资源充足,可轻松扩展更多功能。

85元复刻!开源EDA-Robot 小智AI机器狗,乐鑫ESP32主控,内置DeepSeek、Qwen,可自然语言控制

 

 

 M5 明栈 StackChan  

 

 

图片
StackChan是一款基于乐鑫ESP32-S3物联网芯片的开源AI桌面机器人,内置了AI Agent,支持视频通话、远程遥控智能家居设备,可以陪你玩,陪你聊,还支持开源共创与DIY扩展,配合App与开源生态,你可以把它养成最符合你风格的桌面搭子。
StackChan 集成了红外收发器、三区触摸面板以及全功能 NFC 模块,同时配备了双舵机,支持水平 360° 无限旋转与垂直 90° 俯仰,配合机身内置的 12 颗 RGB LED 灯珠,使得它拥有灵动的眼神、拟人的头部动作和多种表情。
图片
StackChan 基于乐鑫ESP32-S3芯片,其核心驱动力来源于 M5Stack 旗舰级的 CoreS3 开发套件:
  • 主控单元:基于乐鑫 ESP32-S3 SoC,拥有 240 MHz 双核处理性能,支持 Wi-Fi 与蓝牙连接。

  • 视觉感知:内置 30 万像素摄像头与近场传感器

  • 多维交互:配备 2.0 英寸电容式触摸屏、9 轴 IMU、双麦克风及 1W 扬声器。

  • 丰富扩展:具备 microSD 卡槽,支持海量数据存储与离线资源调用。

图片
你只需轻抚背部或触摸头部,它就会抬头用表情回应你,不止屏幕表情,它还有真实的动作反馈:转头、跟随,让你的桌面更有生命。
淘宝预售699元!100%开源,乐鑫StackChan社区共创 AI 桌面机器人,内置AI Agent,支持视频通话、操控智能家居设备

 

 

 小蜘蛛四足 Sesame  

 

 

图片

Sesame Robot 是一个软硬件100%开源的基于乐鑫 ESP32 微控制器,采用 Arduino 技术栈开发的迷你四足小机器人项目,不仅拥有灵活的肢体语言,还通过屏幕赋予了生动的呆萌“表情”,极具亲和力,非常适合 Maker 和机器人爱好者入门。

Sesame Robot 开源仓库包含 CAD 设计文件、STL 打印文件、组装和接线指南,以及基于 ESP32 控制器的基础/扩展固件,如果你有 3D 打印机、熟悉基本焊接技能和 Arduino 编程基础,就能轻松打造一台会走路、会跳舞、拥有“像素脸表情”的可爱机器人。

  • 四足行走:拥有8 个自由度的 MG90S 金属齿轮伺服组成,能实现走路、挥手、跳舞、指向、休息等预设动作。
  • 表情屏幕:128x64 像素的OLED 显示屏作为“表情萌脸”,支持像素风格的表情库:开心、愤怒、害羞等,可随动作同步变化。
  • 外壳骨架:全部使用 PLA 材料3D打印,组装简单,轻薄稳固。

  • 语音交互:通过语音命令控制 Sesame 行走,根据对话实时显示动态表情

图片

 

图片

The EchoEar companion robot features the ESP-Brookesia framework, enabling full-duplex speech interaction, multimodal recognition, and intelligent agent control for a more immersive user experience.

The EchoEar suite emphasizes end-to-end development with the following technical highlights:

Full-duplex voice interaction: Users can engage in continuous dialogue rather than single-shot responses.

Multimodal recognition and agent control: The robot can identify various input modes and provide corresponding intelligent responses.

Emotion recognition and dynamic expression: It detects users' emotional changes and interacts through dynamic expressions, enhancing the device's personality.

Long memory capability: Supports persistent recording of user dialogues for a more companion-like, personalized experience.

Offline voice wakeup and source localization: Combined with motor control, it achieves 180° direction tracking for more natural interactions.

Smart home control: Supports MCP protocol and function call capabilities for remote control of home devices, acting as a local smart home hub.

Lexin open-source EchoEar companion robot connects to Baidu, DeepSeek, Qwen 3, and Wenxin Yiyan.

Tank Man ESP-SparkBot

图片

ESP-SparkBot is a smart device based on the ESP32-S3 development board, integrating voice interaction, image recognition, remote control, and multimedia functions. It enables large-model dialogues (via DeepSeek, OpenAI, or Tongyi), weather queries, music playback, and other interactions using the Xiaodu mobile app for Bluetooth networking, tone switching, music playback, and more.

It also features an accelerometer for games like Shake-the-Dice and 2048. The magnetic design allows modular expansion into a remote-controlled car with camera transmission. The device supports local AI processing for facial and motion detection. Additionally, the projection module enables high-definition video playback and gaming, showcasing its versatility.

Weather clock displays local date, weather, and temperature via IP address.

Large-model AI dialogue connects to DeepSeek, OpenAI, or Tongyi for conversational chat.

Bluetooth speaker with voice control via Xiaodu, app-based volume adjustment, tone switching, and music streaming from QQ Music/Ximalaya.

Voice/face recognition & motion detection runs locally, including cat face and pedestrian detection.

USB projection for use as a secondary monitor, speaker, or microphone, supporting gaming experiences.

Smart car mode with phone/voice control, wireless transmission, and lighting features, controlled via browser interface.

Cyber wooden fish accumulates merit with each tap, with multi-device synchronization increasing merit accumulation.

Casual games include Cyber Dice and 2048.

图片

At only 358 RMB, the ESP32 development board can be turned into a Wall-E robot connecting to DeepSeek and OpenAI.

Xiaozhi EDA-Robot

图片

The EDA-Robot Xiaozhi AI dog utilizes the ESP32-S3R8N16 module combined with INMP441 and MAX98357 for voice dialogue and MCP protocol control. Users can directly issue commands like "move forward" without needing a smartphone app — the robot responds based on natural language understanding.

It includes access to large language models like DeepSeek, Qwen, and Baidu via Xiaozhi, supporting natural language processing.

The MCP protocol allows the LLM to control the robot's movements directly.

Natural language control is fully supported, with no predefined keywords required — the system relies entirely on LLM reasoning.

Custom wake-up word: supports custom naming to wake up the robot dog, creating an exclusive pet.

EDA-Robot XiaoZhi AI robot dog's main control uses the ESP32S3 series module, integrated with WiFi function, providing rich GPIO interfaces. Onboard USB to serial chip for convenient program download and debugging.

The audio acquisition circuit uses the INMP441 digital microphone module for audio acquisition, with a simple structure that can be replaced with a lower cost model as needed.

The audio output part uses the MAX98357 module, supporting I2S audio input and driving speaker output. The design is simple and the performance is stable.

The display screen part uses the SSD1306/SSD1315 OLED screen, with a resolution of 128×64 and a size of 0.96 or 1.3 inches. Pay attention to the I2C line sequence and pull-up resistor configuration.

The power supply circuit adopts two 14500 batteries in series, with a full-charge voltage of 8.4V. Two LDOs are used to step down the voltage to 5V for the servos and 3.3V for the MCU.

The servo circuit uses SG90/MG90 servos, controlled by PWM signals. Since the servos have a built-in 5V to 3.3V drive circuit, the PWM signal can be directly controlled by the MCU IO without level conversion.

The button circuit and download interface are designed to be simple and can be directly connected to the MCU IO. The MCU has sufficient resources to easily extend more functions.

85 yuan replica! Open source EDA-Robot XiaoZhi AI robot dog, using LeXin ESP32 as the main control, built-in DeepSeek, Qwen, and can be controlled by natural language.

M5 Mingzhen StackChan

图片

StackChan is an open-source AI desktop robot based on the LeXin ESP32-S3 IoT chip, with a built-in AI Agent, supporting video calls, remote control of smart home devices, and can accompany you to play and chat. It also supports open-source co-creation and DIY expansion, and can be cultivated into a desktop companion with your own style through the App and open-source ecosystem.

StackChan integrates infrared transceiver, three-zone touch panel, and full-function NFC module. It is also equipped with dual servos, supporting 360° horizontal infinite rotation and 90° vertical pitching. Combined with the 12 RGB LED beads on the body, it has dynamic eyes, anthropomorphic head movements, and multiple expressions.

图片

StackChan is based on the LeXin ESP32-S3 chip, and its core driving force comes from M5Stack's flagship CoreS3 development kit:

Main control unit: Based on the LeXin ESP32-S3 SoC, with 240 MHz dual-core processing performance, supporting Wi-Fi and Bluetooth connections.

Visual perception: Built-in 300,000-pixel camera and proximity sensor

Multi-dimensional interaction: Equipped with a 2.0-inch capacitive touch screen, 9-axis IMU, dual microphones, and a 1W speaker.

Rich expansion: Has a microSD card slot, supporting massive data storage and offline resource calling.

图片

You only need to gently touch the back or head, and it will look up and respond to you with expressions. It not only has screen expressions but also real action feedback: turning its head and following, making your desk more alive.

Pre-sale on Taobao for 699 yuan! 100% open source, LeXin StackChan community co-created AI desktop robot, built-in AI Agent, supporting video calls and control of smart home devices.

Small spider four-legged Sesame

图片

Sesame Robot is a 100% open-source hardware and software project based on the LeXin ESP32 microcontroller, using the Arduino technology stack. It is a mini four-legged robot that not only has flexible body language but also has a vivid "expression" screen, making it highly approachable and very suitable for makers and robot enthusiasts to get started.

The Sesame Robot open-source repository contains CAD design files, STL printing files, assembly and wiring guides, and basic/extended firmware based on the ESP32 controller. If you have a 3D printer, are familiar with basic welding skills, and have Arduino programming basics, you can easily build a cute robot that walks, dances, and has "pixel face expressions".

Four-legged walking: It consists of 8 degrees of freedom MG90S metal gear servos, which can perform preset actions such as walking, waving, dancing, pointing, and resting.

Expression screen: A 128x64 pixel OLED display as a "facial expression" screen, supporting pixel-style emoticon libraries: happy, angry, shy, etc., which can change synchronously with the action.

Shell skeleton: All 3D-printed with PLA material, simple to assemble, light and stable.

Voice interaction: The robot can be controlled by voice commands to walk, and it will display dynamic expressions in real-time according to the conversation.

图片

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry