Alibaba AI sets 16 records, beats Google and NVIDIA in robotics

Alibaba has introduced RynnBrain, a new artificial intelligence system designed to enhance robots' understanding of space, objects, and motion in real-world environments. This model represents Alibaba’s significant push into "physical AI," a field focused on embodied AI systems such as robots and autonomous vehicles that interact directly with their surroundings. RynnBrain addresses key limitations in existing robotics AI by incorporating spatiotemporal memory, enabling robots to remember object locations over time and predict their movements. It also supports global retrospection, allowing robots to review past actions to reduce errors, and combines text-based logic with spatial reasoning for more accurate real-world interaction. The system features a 30-billion-parameter mixture-of-experts model but activates only 3 billion parameters during inference, leading to greater efficiency and faster robot decision-making compared to larger models like Google’s Gemini Robotics ER 1.5 and NVIDIA’s Cosmos Reason 2. Alibaba reports that RynnBrain set 16 new records on open-source embodied AI benchmarks, excelling