Alibaba introduced an open-source artificial-intelligence model, dubbed RynnBrain, to power robots that can recognize and manipulate real-world objects—advancing China’s bid to lead in “physical AI.” A demo from Alibaba’s DAMO Academy shows a robot sorting fruit, a seemingly simple task that hinges on complex perception and control. The move positions Alibaba alongside Nvidia, Google’s DeepMind and Tesla, which are building “world models” and platforms for embodied AI, including humanoids. RynnBrain extends Alibaba’s Qwen family of models and is intended to spur developer adoption globally through open sourcing. The effort underscores intensifying U.S.-China competition in robotics and autonomy as companies race to commercialize factory, logistics and service applications.
Related articles:
ROS.org | Robot Operating System
Mobile ALOHA: Learning bimanual mobile manipulation for household tasks
OpenVLA: An open-source vision-language-action model for robotics
