Kodro turns robotics learning into one visual loop: Design → Code → Simulate → Prove → Build. Students can assemble robots, program them with Python or blocks, test them in interactive 3D worlds, inspect live sensors and export evidence, all offline-first. Teachers get 24 curriculum lessons plus a lesson creator, while makers can prototype without expensive hardware. No account, cloud connection or AI is required.
Hey Product Hunt! I’m Vaibhav, the maker of Kodro.
I built Kodro because beginner robot programming can feel unnecessarily inaccessible: physical kits cost money, setup is fragile, and many simulators are built for professional robotics workflows rather than pupils or first-time coders.
Kodro is a free, open-source and offline-first place to design a robot, write Python, and watch it attempt a task in a visual world.
You can:
• Learn with 24 guided coding lessons, hints and worked solutions
• Build a rover from hobby parts and see how choices affect range, speed and stopping distance
• Program through a beginner-friendly Python subset or blocks
• Test ideas in a room, city, deserts, underwater environments, Mars and space
• Create and share your own class lessons with deterministic grading
The important constraint is that the core experience works without an account, API key or internet connection. Optional AI help can run locally through Ollama, while a deterministic offline fallback remains available when no model is installed.
Kodro is an MSc research project from the University of Liverpool, and I’d really value feedback from educators, parents, makers and developers:
What would make a robot-coding simulator genuinely useful in your classroom, club or learning journey?
Try it here: https://vaibhav4046.github.io/ko...
Source code: https://github.com/vaibhav4046/k...
Report
I could see this being useful before spending money on a physical kit. Are you planning to connect it with real hardware too?
@sheikh_umair1 Yeah that is the future plan, as it was my collage project and didn't have enough resources around me to make the hardware connection possible, but soon i will work on update of kodro which will have significant changes then previous and hopefuly hardware connection i will work on.
Report
Nice robotics angle. Can you prove a robot offline and drop the same code onto real hardware without rewriting?
the MSc research project background explains why the deterministic grading and offline-first constraints feel so deliberate rather than bolted on. curious how the Ollama-backed AI help degrades - does the offline fallback still give useful hints, or is it a much thinner experience without a local model installed?
@vaibhav4046 good to hear it's not a stripped-down mode. so the deterministic offline logic actually carries most of the pedagogical weight and the local model is more of a bonus layer on top, not a crutch the lessons depend on - that's a healthier design than I assumed going in.
Kodro
I could see this being useful before spending money on a physical kit. Are you planning to connect it with real hardware too?
Kodro
@sheikh_umair1 Yeah that is the future plan, as it was my collage project and didn't have enough resources around me to make the hardware connection possible, but soon i will work on update of kodro which will have significant changes then previous and hopefuly hardware connection i will work on.
Nice robotics angle. Can you prove a robot offline and drop the same code onto real hardware without rewriting?
Dial
the MSc research project background explains why the deterministic grading and offline-first constraints feel so deliberate rather than bolted on. curious how the Ollama-backed AI help degrades - does the offline fallback still give useful hints, or is it a much thinner experience without a local model installed?
Kodro
@galdayan you can run it fully offline and it is not thinner experience without local model, you can absolutely use it without it.
Dial
@vaibhav4046 good to hear it's not a stripped-down mode. so the deterministic offline logic actually carries most of the pedagogical weight and the local model is more of a bonus layer on top, not a crutch the lessons depend on - that's a healthier design than I assumed going in.