Creabo—DIY robot kit powered by LLM

What it does

Creabo, an AI-powered DIY robot kit. Users can build and program robots using natural language making robotics accessible to all. It lowers entry barriers, boosts efficiency, promotes inclusive innovation through the idea "everyone designs, everyone creates."

Your inspiration

Traditional DIY kits face high technical barriers, limiting accessibility and creativity. AI large language models offer a breakthrough: they understand natural language, generate code, guide users, and help robots learn by analogy—greatly simplifying programming and expanding capabilities. This opens the door to a new type of DIY kit—one that integrates AI for natural language control, making robotics more intuitive, scalable, and user-friendly.

How it works

Inspired by the powerful knowledge base and generalized understanding capabilities of large language models, we have designed and implemented a complete AI-driven DIY robotics kit powered by large language models. After pre-training, the large language model can fully understand the context of users' DIY robot creation, providing detailed guidance during debugging and assembly to reduce the difficulty of the development process. Pre-set kit packages and a collaborative community provide DIYers with creative inspiration, allowing users to unleash their creativity, share experiences, and enhance the freedom, collaborative nature, and overall experience of the robot DIY process.

Design process

We began by analyzing the strengths of AI large language models—natural language interaction, strong generalization, and a vast knowledge base—and derived corresponding design principles for DIY robot kits: precise output control, rich expandability, simple and reliable interfaces, and flexible connection structures. Guided by this, we iterated through five design stages: from a basic single-motor unit, to an expandable version with slots, then a mobility-focused "big-footed" design, which was later split into Z-shaped modules for enhanced control and modularity. Finally, we unified fixed/movable interfaces on Z-modules to support diverse DIY forms. Hardware was also optimized: lithium batteries with voltage boosting reduced size and improved performance; internal layout and wiring were streamlined; control software adopted zero-current control for heat reduction and better battery life; micro-motion programs enhanced motor precision and positioning. For AI training, we used a large language model API connected via Wi-Fi/Bluetooth to a microphone-equipped control board. By standardizing data formats and configuring initial tools and actions, the AI achieved effective "learning by analogy," enabling more natural, intelligent robot behavior.

How it is different

Traditional DIY kits often cater to niche communities, with high learning curves and complex intent-to-command translation. In contrast, this robotics kit leverages large language models to simplify the entire development process. In the creative phase, pre-set kits, communities, and LLM offer inspiration, boosting engagement and product longevity. During robot construction, the LLM enables low-code or no-code development. In debugging and assembly, it provides step-by-step guidance, greatly lowering the barrier to entry. Most importantly, it interprets user intent through natural dialogue and translates it into machine instructions, streamlining interaction. This approach broadens the user base: for children, it's a creative toy; for home users, an intelligent assistant; for makers, a flexible toolkit; for researchers, a prototyping platform. Regardless of technical background, the LLM acts as a smart core, making the kit intuitive, open, and engaging for all

Future plans

Goal 1: Create ready-to-use accessory kits for both entertainment and functionality to inspire users and broaden adoption. Goal 2: Enhance the language model’s creativity, accuracy, and personalized support. Goal 3: Improve interaction flow by better integrating LLM guidance into user creation and execution phases. Goal 4: Partially open-source the project, commercialize the product, and build a user-driven community to foster sharing and market growth.

Awards