Max Robotics — Global Service Robot Directory + US Certification

About

Stefano Saliceti is a leading robotics researcher whose work sits at the intersection of deep reinforcement learning, agile locomotion, and whole-body manipulation. He is best known for demonstrating that deep RL can synthesize sophisticated, safe movement skills for low-cost humanoid robots. His landmark 2024 paper, *Learning Agile Soccer Skills for a Bipedal Robot with Deep Reinforcement Learning* (147 citations), showed a miniature humanoid learning to play one-versus-one soccer—a feat requiring dynamic balance, rapid decision-making, and coordinated kicking. This work proved that complex, real-time athletic behaviors are achievable on affordable hardware. Saliceti has also pushed beyond simple pick-and-place tasks, tackling robotic stacking of diverse, irregularly shaped objects using vision-based RL. His contributions to quadruped agility include co-developing the Barkour benchmark (2023), which set a standard for measuring animal-level locomotion skills like sprinting and leaping. More recently, he contributed to Google DeepMind’s *Gemini Robotics* (2025), a family of multimodal AI models designed to bring generalist intelligence into the physical world. Across his career, Saliceti has consistently advanced the frontier of real-world robot learning, from telemanipulation with haptic feedback to wild-environment bipedal locomotion.