On November 6, the finalist teams for the first Ant Lingbo Embodied Large Model Challenge were officially announced: 15 teams and 59 contestants stood out from 1,855 teams and 3,252 contestants worldwide, securing their places in the offline finals.
(Photo: Finalist list for the first Ant Lingbo Embodied Large Model Challenge)
Top 15 Assemble for a 48-Hour Real-Robot Hackathon Showdown
This year's Challenge focuses on the interaction capabilities between embodied large models and the physical world, attracting participants from 13 countries, 648 universities, research institutes, technology companies, and embodied intelligence teams worldwide. According to AMD, one of the competition's computing power partners, the number of GPUs used for the preliminary round increased from the originally planned 200 to 350, providing more than 340,000 GPU-hours of computing support across the two stages. The scale of participation and computing investment also reflect the current enthusiasm across industry and academia for exploring embodied large models.
Using preliminary-round task success rate as the main ranking criterion, the Challenge conducted re-evaluations of top-performing teams and ultimately determined the top 15. The advancing contestants come from universities such as Tsinghua University, Peking University, Fudan University, and Shanghai Jiao Tong University; research institutes including Shanghai AI Laboratory and relevant institutes under the Chinese Academy of Sciences; and companies such as Alibaba, Kuaishou, and Pudu Robotics, bringing together strengths in algorithm research and robot software and hardware engineering.