A fully open-source AI robot featuring natural conversation, autonomous navigation, computer vision, and a modular 3D-printed body.
Millie Bot turns your home into a modern technology hub
Millie Bot is an open-source AI companion robot designed to make personal robotics more accessible than ever before. Built around natural conversation, autonomous navigation, and a fully 3D-printed design, Millie combines modern artificial intelligence with an approachable platform that anyone can explore, customize, and improve. Whether you're a developer, maker, student, or simply curious about the future of robotics, Millie is designed to be a project you can learn from and grow with.
At its core, Millie is more than just a robot. It is a flexible robotics platform capable of navigating its environment, recognizing people, completing tasks, and interacting through natural voice conversations. Powered by ROS2 and built with open-source software and hardware, Millie encourages experimentation and gives the community the freedom to create new features, personalities, accessories, and experiences. Every improvement has the potential to benefit builders around the world.
Millie Bot is an evolving community project where innovation is driven by collaboration. As development continues, new capabilities, hardware designs, software improvements, and educational resources will be shared openly so others can build upon them. By bringing together robotics, artificial intelligence, and an open community of creators, Millie aims to make personal robotics more accessible while inspiring the next generation of builders to imagine what a companion robot can become.
Millie Bot is designed to interact with the world in a natural and meaningful way. By combining artificial intelligence, autonomous navigation, computer vision, and environmental awareness, Millie can hold conversations, move independently, recognize people, and respond to its surroundings in real time. These capabilities work together to create a companion robot that can explore its environment, communicate naturally, and accomplish simple tasks.
Millie is built as a flexible platform that continues to evolve through ongoing development and community contributions. From AI voice conversations and autonomous navigation to human detection, computer vision, task automation, and remote control, each capability is designed to work seamlessly with the others. The following sections explore the core technologies that power Millie and how they work together to create an intelligent, open-source companion robot.
Millie Bot uses the OpenAI Realtime API to deliver natural voice conversations through two distinct interaction modes. Realtime Streaming allows Millie to listen and respond continuously with low latency, creating a fluid conversation where responses begin almost immediately as ideas are understood. Turn Taking provides a more traditional experience, where Millie waits for you to finish speaking before generating a complete response. Together, these modes allow conversations to be tailored for everything from casual discussions to more deliberate question-and-answer interactions.
OpenAI's advanced voice models provide a growing collection of expressive, high-quality voices that give Millie a unique personality. Whether you prefer a calm assistant, an energetic companion, or another speaking style, voices can be changed to match the experience you want. Combined with natural language understanding and conversational memory, Millie is designed to feel approachable, engaging, and enjoyable to interact with over time.
The AI voice agent is also deeply integrated with the rest of the robot through a powerful tool system. Rather than simply answering questions, the AI can control Millie's movement, navigate to locations, initiate autonomous behaviors, manage tasks, and interact with other onboard systems using natural language. This allows spoken conversations to become real-world actions, transforming Millie from a voice assistant into an intelligent companion capable of understanding requests and carrying them out in the physical world.
Source: https://developers.openai.com/api/reference/resources/realtime
Millie Bot is built on ROS2, an industry-standard robotics framework that provides the foundation for communication between the robot's software, sensors, and hardware. Using the Navigation2 (Nav2) stack, Millie can plan efficient paths, avoid obstacles, and navigate complex indoor environments without requiring constant user input. This modular architecture makes it easy to expand the robot with new capabilities while benefiting from one of the largest open-source robotics communities in the world.
To understand its surroundings, Millie uses LiDAR together with the SLAM Toolbox to create detailed maps of the spaces it explores. During Mapping Mode, the robot continuously scans its environment, allowing users to generate, save, and reuse maps of their home or workspace. Once a map has been created, Millie can accurately localize itself within that environment and confidently navigate between rooms while adapting to changes and avoiding obstacles.
Autonomous navigation is tightly integrated with the rest of Millie's AI system. Voice commands, task automation, and future behaviors all build upon this navigation framework, allowing the robot to travel to specific locations, follow people, patrol mapped areas, or complete simple tasks independently. By combining ROS2, Nav2, LiDAR, and SLAM Toolbox, Millie gains the awareness and mobility needed to interact naturally with the physical world.
Millie Bot can intelligently follow people throughout a mapped environment using a combination of human detection, autonomous navigation, and real-time path planning. Once a person has been identified, the robot continuously tracks their movement while navigating around obstacles and adjusting its path as they move. This creates a natural companion experience, allowing Millie to accompany users from room to room without requiring manual control.
Person Following can be activated in several ways depending on how the robot is being used. Users can start the mode manually through a voice command or the controller tablet, allowing Millie to immediately begin following them. In Stationary Tracking mode, the robot remains in place while watching for movement, automatically transitioning into person following if the user walks away. During Patrol mode, Millie autonomously explores a mapped environment and can react when someone is detected, pausing its patrol to greet the user, begin a conversation, or follow them as they continue through the space.
Source: https://docs.ros.org/
Source: https://docs.nav2.org/
Source: https://docs.ros.org/en/ros2_packages/humble/api/slam_toolbox/
Included with Millie Bot is Millie Mini, a stationary desktop AI agent that runs on the controller tablet and brings Millie's personality to life wherever you go. Housed inside a compact 3D-printed body modeled after the full-size robot, Millie Mini features the same animated face, AI voice agent, and conversational experience in a desktop form factor that fits comfortably on a desk, shelf, or workspace. It provides a fun and approachable way to interact with Millie throughout the day while serving as a familiar companion wherever you are.
Millie Mini is also deeply integrated with the controller application that powers the full-size robot. The controller app provides access to AI conversations, live camera feeds, LiDAR mapping, navigation controls, task management, robot settings, and remote operation, all from a single interface. Because Millie Mini runs on the same tablet, users can seamlessly switch between talking with their AI companion and managing the robot, creating a consistent experience whether Millie Bot is nearby or operating elsewhere.
Together, Millie Mini and the controller application create an ecosystem that extends beyond the robot itself. They allow users to continue conversations, develop new features, test AI personalities, and stay connected to the platform even when the full-size robot is not in use. As the Millie Bot project continues to evolve, Millie Mini will grow alongside it, bringing new AI capabilities and helping create a more connected and accessible companion robotics experience.
Millie Bot is built around a modular hardware platform designed for reliability, expandability, and ease of development. At the heart of the robot is a compact Mini PC that runs the core robotics software, including ROS2, autonomous navigation, computer vision, sensor integration, mapping, and communication between the robot's various systems. Two Android tablets complement the platform, with one serving as Millie's expressive animated face while also connecting to the OpenAI API for AI voice conversations. The second tablet functions as a portable controller, providing live camera feeds, LiDAR mapping, navigation, task management, robot settings, and remote control.
Low-level motion control is handled by an Arduino Nano Every, which interfaces with the Mini PC while controlling the robot's two motor drivers. Millie is powered by a 12V LiFePO₄ battery, with a dedicated fuse box protecting each electrical circuit. A 19V step-up converter supplies power to the Mini PC, while 5V step-down converters are used to charge the face tablet and provide power to two USB charging ports. DC motors provide smooth differential drive movement, and integrated cooling fans help maintain reliable operating temperatures during extended operation.
A TP-Link AX 1500 WiFi 6 Travel Router creates a dedicated private network that connects the Mini PC, Android tablets, and onboard devices while providing internet connectivity for the OpenAI API and other online services. This architecture keeps communication fast and reliable between every component of the robot while making it easy to integrate additional sensors, cameras, actuators, and future hardware as the Millie Bot platform continues to evolve.
Source: https://docs.arduino.cc/hardware/nano-every/
Source: https://www.tp-link.com/us/home-networking/wifi-router/tl-wr1512x/
Millie Bot is built as a fully open-source robotics platform, making it easy for developers, makers, students, and robotics enthusiasts to learn from, customize, and improve. Rather than treating the robot as a closed product, the project is designed to encourage collaboration through openly available software, hardware designs, and documentation. Every part of Millie is intended to be explored, modified, and shared, allowing the community to contribute new ideas while helping the platform continue to evolve.
The open-source ecosystem includes the robot's source code, a complete FreeCAD assembly model, and downloadable 3D printing files for the robot's components. Together, these resources provide everything needed to understand how Millie works, build your own robot, customize its appearance, develop new features, or contribute improvements back to the community. By making both the software and hardware openly available, Millie Bot aims to create a collaborative robotics platform where innovation is driven by everyone who participates.
Millie Bot's software is developed primarily in Python and Flutter, creating a flexible platform that is easy to understand, customize, and expand. Python powers the robot's core systems, including artificial intelligence, autonomous navigation, computer vision, voice interaction, and communication between software components. Flutter is used to develop both Millie's animated face and the controller application, providing modern, cross-platform user interfaces that are responsive, intuitive, and easy to extend with new features.
Because the software is fully open source, developers have complete freedom to build upon the platform and shape its future. New AI voice capabilities, personalities, facial expressions, animations, and user interfaces can be added alongside support for new sensors, robotics packages, and hardware configurations. The modular architecture also makes it possible to develop entirely new applications and capabilities, from robotic grippers and home automation to specialized tools and behaviors, creating a platform that continues to evolve through community collaboration and advances in artificial intelligence and robotics.
Source: https://www.python.org/
Source: https://flutter.dev/
Millie Bot was designed in FreeCAD, a free and open-source parametric CAD application that makes mechanical design accessible to anyone. The complete robot is provided as a fully editable assembly model, allowing builders to explore how every component fits together, inspect the mechanical design, and modify individual parts to suit their own projects. Because the model is parametric, dimensions and features can be adjusted without starting from scratch, making it easy to customize the robot while maintaining a clean, organized design. FreeCAD also includes powerful assembly tools, exploded views, and bill of materials support, making it an excellent platform for learning robotics design as well as maintaining and improving the project.
By sharing the complete FreeCAD assembly, Millie Bot becomes more than a finished robot—it becomes a platform for experimentation and innovation. Builders can redesign body panels, create custom mounts, integrate new sensors, modify the chassis, or adapt the robot for entirely new applications while using the original design as a foundation. Whether you're learning mechanical design for the first time or contributing improvements back to the community, FreeCAD provides a professional, feature-rich tool that keeps the entire hardware platform open, accessible, and ready for future development.
Millie Bot features a fully 3D-printed modular design that makes building, repairing, and customizing the robot both practical and accessible. The body is divided into individual components that can be printed, assembled, and replaced independently, allowing builders to repair damaged parts or upgrade specific sections without rebuilding the entire robot. This modular approach also simplifies assembly and encourages experimentation with new designs as the project continues to evolve.
Every printable component is provided as part of the open-source project, giving the community the freedom to modify body panels, create custom accessories, redesign mounting systems, or adapt the robot for specialized applications. Whether you're adding new sensors, experimenting with different wheel configurations, designing a robotic gripper, or simply giving Millie a unique appearance, the modular design makes customization straightforward while maintaining compatibility with the rest of the platform.
Millie Bot is more than a robotics project. It is an opportunity for people to collaborate on a shared platform that continues to grow through creativity, experimentation, and innovation. Every new feature, design improvement, or idea has the potential to make the robot more capable while creating something that the entire community can build, enjoy, and learn from together.
Some of the most impactful technologies in the world have been shaped by communities working toward a common goal. Millie Bot embraces that same spirit by providing a platform where people can explore new ideas, develop new skills, and contribute to a product that becomes more capable with every contribution. By working together, we can make robotics more accessible, inspire future innovation, and create technology that benefits everyone.
Millie Bot is one project within a much larger vision. Dream Cloud exists to bring together people who want to build technologies, businesses, communities, and ideas that create a better future. Whether your passion is robotics, artificial intelligence, education, healthcare, sustainability, or something entirely different, there is a place for your ideas within that vision.
Building the Dream World is an open invitation to create something meaningful together. We believe the future will be shaped by people who are willing to share their knowledge, support one another, and work toward common goals. If Millie Bot inspires you to start a project, contribute to open source, or launch an idea of your own, then it has already accomplished something far greater than building a robot. It has helped grow a community of people committed to building a better future together.
OpenDroids is developing and deploying adaptable robotic systems for hospitality, logistics, manufacturing, retail, and other industries, combining custom hardware, AI, integration, and Robotics-as-a-Service.
Reflex Robotics is deploying general-purpose humanoid robots into logistics and industrial environments, using real-world work and human supervision to help its systems become increasingly capable.
Tesla could combine existing automated food-production systems with Optimus to create a real-world environment where humanoid robots learn the physical tasks that connect an autonomous restaurant.
Building the integration layer that connects specialized food-service robotics into one modular, autonomous operating system.
Developing highly automated food production facilities that combine robotics, industry expertise, and regional partnerships to reduce costs and supply prepared food at commercial scale.
Creating real-world deployment opportunities where robots can perform practical tasks and generate valuable training data for physical AI.
