Chinese team wins top prize in global robotics grasping contest

  • Victory highlights advances in embodied AI for cluttered, real-world environments
  • FAM model demonstrates high reliability with minimal training data

A Chinese team captured the top spot in a global robotics competition focused on object manipulation, a capability widely viewed as critical for deploying embodied AI in warehouses and factories.

The win came at the 2026 IEEE International Conference on Robotics and Automation (ICRA) in Vienna, where the Youth2Real team competed in the Picking in Clutter Track of the 11th Robotic Grasping and Manipulation Competition (RGMC).

The team consisted of members from Hangzhou-based embodied robotics startup Five Ages (中科第五纪) and researchers led by Tsinghua University professor Sun Fuchun.

Ultra-low sample

The contest replicates complex, real-world scenarios in warehouses and industrial settings, with randomly stacked objects in varying orientations. Robots must autonomously identify, locate, plan, and execute grasping tasks.

Five Ages’ FAM series of ultra-low-sample embodied AI models powered the team’s robot.

Unlike traditional approaches requiring massive datasets, FAM employs “heatmap alignment” to efficiently capture 3D spatial features from 2D inputs, greatly reducing sample dependence.

The technology had previously won the CVPR 2025 Embodied Manipulation Competition and earlier this year saw its next-generation world model, FlowWAM, top the global WorldArena leaderboard.

International validation

Founded in September 2024 by Liu Nianfeng (刘年丰), a PhD from the Chinese Academy of Sciences, Five Ages has attracted multiple rounds of funding from leading investors including HongShan and Oriental Fortune Capital.

“Robot operational capability is a key step for embodied AI to achieve industrial deployment,” a Five Ages spokesperson said.

The international victory validates the feasibility of a “low-sample, high-reliability, strong-generalization” approach and demonstrates the practical deployment potential of China’s embodied AI technology in logistics and industrial manufacturing, the spokesperson added.