Scientists taught an AI-powered 'robot dog' how to play badminton against humans — and it's actually really good

BLUF

Scientists have trained the quadruped robot ANYmal to play badminton using AI and reinforcement learning, demonstrating advanced coordination of vision, limb movement, and decision-making, with potential applications in dynamic task performance beyond sports

Learning Outcomes

• Understand the integration of AI, reinforcement learning, and multi-joint coordination in autonomous robotics.
• Evaluate how dynamic visual perception and locomotion control can be applied to real-world operational environments.
• Explore potential uses of agile AI-powered robots for disaster relief, logistics, and complex terrain navigation.