A Swedish AI startup has developed a robot dog named Luna, equipped with a functional digital nervous system that enables it to learn and adapt like humans and many animals. IntuiCell describes Luna as one of the first applications of physical agentic AI, capable of making decisions and taking actions toward specific goals rather than performing narrow tasks or generating content.
Instead of incorporating generative AI models and extensive datasets, IntuiCell plans to hire a dog trainer to teach Luna to walk, simulating the way neurons interact and process information. CEO and co-founder Viktor Luthman stated, "What we have built is the first software that allows any machine to learn as humans and animals do," emphasizing that there’s no need for pre-training, offline simulations, or expansive data centers.
Luthman believes that the robot dog's capabilities could significantly enhance the development of human-like robots in various environments. He envisions the next steps involving humanoid and autonomous robotics for unpredictable settings such as space exploration, deep-sea missions, or disaster response. Intelligent machines could, for example, be deployed to Mars to construct future habitats, operating without prior training and addressing problems as they arise.
Currently, Luna can stand on its feet and is designed to perceive, process, and improve through direct interactions with its surroundings. Established in 2020 as a spin-out from Lund University, IntuiCell is backed by Sweden's Navigare Ventures, Norway's SNÖ Ventures, and the European Union.