Swedish startup develops robot with learning 'digital nervous system'

IntuiCell, a deep-tech startup from Stockholm, has created a unique AI system that learns like a biological nervous system. This technology was demonstrated on a quadruped robot that can learn in real time without needing pre-programmed instructions. The company's CEO, Viktor Luthman, explained that traditional AI has struggled to achieve true intelligence. IntuiCell's system allows machines to evolve and adapt through interaction, much like animals learn from their trainers. To showcase this, IntuiCell plans to hire a dog trainer to teach the robot new skills. Unlike traditional AI models that rely on fixed datasets, IntuiCell's robot dog, named Luna, learns by interacting with its environment. Luna has already shown that it can control its body and learn to stand through trial and error. IntuiCell aims to create a full digital nervous system within two years. This system will enable AI to learn dynamically and adapt to new challenges without large datasets. The goal is for their technology to help agents—both physical and digital—learn continuously throughout their lives. Many existing AI systems rely on backpropagation and large amounts of training data, which limits their learning capabilities. IntuiCell believes its approach can lead to advances in robotics and industrial automation, allowing AI to tackle complex, real-world problems. Other companies working in this field include Sanctuary AI, Neura Robotics, Google DeepMind, and Boston Dynamics. Upcoming events like the 2025 Robotics Summit & Expo will explore the future of AI and robotics.