
From articulating hands and feeling skin to listening and learning, the amount of problems to be solved in a humanoid robot is so vast it makes you wonder why anybody would even try. A casually friendly sci-fi robot always seems to be just over the horizon, but Christoph Kohstall’s work is the best argument I’ve seen that it might come true. His robot Mona won a blue ribbon at Maker Faire Bay Area in 2023, and he came back in 2024 to share the process and inspire future robot builders.
Christoph Kohstall built an intelligent humanoid robot with human-level motion capabilities entirely from scratch. The robot leverages a large language model to decide how to act in unstructured environments. The goal is everything from working in home care all the way to assisting humans to build an inspiring tomorrow.
He opens up the talk with an interactive demo exploring what robot people actually want in their homes. While the answer was a little more evil than I expected, it was still very much humanoid. So if you want to build tools to live with people, and you want folks to be comfortable around them, then a humanoid form is a sensible fit. There’s always danger lurking in the uncanny valley, but fortunately makers are working to build friendly, interactive robotic humanoids.
And, if you built a robot that had a humanoid form and was generally useful for different things, then you’d probably want it to learn and respond naturally. So he integrated the latest multi-modal LLMs to train not just from text, but data like video too.
In his talk at the fair, Cristoph demonstrated the philosophy behind the robot and showed off some of the capabilities. He even gets his family in on the demonstration.
The build deliberately uses budget parts with the goal of making assistive humanoid robots more accessible. The total BOM is still a lot; he is trying to build a life-size person after all. But the results speak for themselves.
If you’re wondering where the name Kind Humanoid came from, Christoph’s physics background gave him a unique perspective on the potential dangers of new technology (not the first time someone has worried about that), so he wanted to focus on the positive. As he put it, he gave the talk “to motivate people to work towards the inspiring side of it and not get sucked into the wrong kind of application just because it pays well enough.”
This talk changed my idea of what the limits on machines can be, and inspired hope in the team making it happen.