Back to feed
News Story
APriority77
机器之心
1 sources

Lingxi Zhiyong's ROSS Harness Secures Top Three in Robot Olympics Industrial Category

At the 2026 World Humanoid Robot Olympics industrial scenario competition, Lingxi Zhiyong's robot completed assembly and feeding tasks fully autonomously, scoring 160 points and ranking in the national top three, making it the only startup to achieve this. Founded by Duan Yifan, the company combines the CONWAY model with the ROSS Harness to address challenges in industrial embodied intelligence, including success rate, throughput, cost, and migration.

SynthePulse Insight · AI deep readingMembers

From 'Can the Model Do It?' to 'Can the System Deliver?': ROSS Harness and the Second Test for Industrial Embodied AI

Version 1 · 1 source

At the 2026 World Humanoid Robot Games industrial scenario competition, Lingxi Zhiyong secured third place nationally with 160 points, driven by its dual-layer approach of 'CONWAY model + ROSS Harness.' This article dissects the technical logic behind this victory: while the industry is still competing on model capabilities, what industrial sites truly need is the ability to convert models into stable, deliverable system capabilities.

  • Lingxi Zhiyong secured third place nationally with a total score of 160 points in the 'Assembly and Feeding Station' event at the 2026 World Humanoid Robot Games industrial scenario competition, making it the only startup to enter the top three in its debut.
  • The competition required completing three consecutive tasks: transporting 10kg material boxes, picking seven types of parts, and sub-millimeter assembly of 18 engine valves, with scores of 84, 65, and 21 points respectively, plus a 10-point special deduction.
  • Lingxi Zhiyong's technical route is 'CONWAY industrial native model + ROSS Harness autonomous evolution engine,' emphasizing that the model handles 'how to do it' while the Harness ensures 'getting the job done well.'
Open section navigationCompetition Results: Tasks and Challenges Behind the 160 Points

Competition Results: Tasks and Challenges Behind the 160 Points

At the 'Assembly and Feeding Station' event of the 2026 World Humanoid Robot Games industrial scenario competition, Lingxi Zhiyong's robot had to complete three consecutive tasks: transporting 10kg material boxes, picking seven types of parts, and sub-millimeter assembly of 18 engine valves. It finished with scores of 84, 65, and 21 points, and after a 10-point special deduction, a total of 160 points, ranking third nationally among enterprises, behind two teams from leading industry companies, making it the only startup to enter the top three in its debut.

The significance of this competition lies in its encapsulation of real industrial embodied AI problems: not only must the robot 'know how to do' in real scenarios, but it must also do it correctly within industrial time windows, stringing multiple actions into a stable, complete task, and producing deliverable, acceptable industrial results.

Free for now

Read the full analysis

4 more sections of analysis, plus the full takeaway

Loading

Credibility boundary

This article's information primarily comes from Machine Intelligence's report on Lingxi Zhiyong, which is promotional in nature. Some data (such as the 3.6x improvement) lacks third-party verification and should be considered self-reported. Competition scores and rankings are as reported and have not been officially confirmed.

Sources for this version

  1. 跻身世界人形机器人运动会工业场景赛全国前三，ROSS Harness 以全自主模式突破工业具身智能落地瓶颈机器之心

Primary report

机器之心

Primary source