More than 1,800 teams from 13 countries have registered for the inaugural Robbyant Embodied AI Challenge, an international competition focused on developing and evaluating artificial intelligence models for robotic applications. The competition will conclude with a real-robot hackathon in Shanghai from November 14 to 16, 2026.
Organized by Robbyant, an embodied AI company within Ant Group, in partnership with ModelScope and Alibaba Cloud Tianchi, the competition was launched in September at the 2026 Inclusion Conference on the Bund. The online qualification round closed on October 27.
The challenge is based on LingBot-VLA 2.0, an open-source embodied foundation model, and evaluates participants’ capabilities in model training and fine-tuning, generalization in simulated environments, and adaptation to physical robotic systems.
The competition has drawn thousands of developers from universities, research institutions, technology companies and robotics startups. Participating Chinese institutions include Tsinghua University, Peking University, Shanghai Jiao Tong University and Zhejiang University.
International academic participants include teams from Carnegie Mellon University, Georgia Institute of Technology, University College London, Technical University of Munich, National University of Singapore and Nanyang Technological University.
Developers from technology and industrial companies, including Tencent, Alibaba, ByteDance, Meituan, Microsoft, BYD and Xiaomi, have also registered. Robotics companies represented in the competition include UBTECH, Fourier Intelligence and Agibot.
According to registration data provided by the organizers, universities and research institutions account for 74.7% of participating teams, while master’s and doctoral students represent 67.4% of participants. Developers from enterprises and engineering roles account for 25.2%, and 51.7% of participants have backgrounds in vision-language-action (VLA) systems or multimodal foundation models.
The competition provides a common evaluation environment for developers to train and test embodied AI models across simulated and physical robotic tasks. Its final stage in Shanghai will focus on applying the participating teams’ models to real robots, following the online qualification process.
Through access to LingBot-VLA 2.0 and associated evaluation environments, the organizers aim to support experimentation with embodied foundation models and their application to robotic systems.
