In the quiet corridors of the Qilu Software Park in Jinan, the usual morning rush of office workers is now accompanied by an unusual rhythmic cadence. This proves not the sound of a dedicated marathoner’s sneakers, but the steady, mechanical stride of a humanoid robot pushing its limits in a grueling regimen of endurance training.
This robotic athlete, known as “Walker Taishan,” is currently engaged in what its engineers call a “devil’s training” cycle. Since early March, the machine has been tasked with running a full half-marathon distance—21.0975 kilometers—every single day. The goal is not merely to survive the distance, but to prepare for one of the most ambitious intersections of athletics and technology: the 2026 Humanoid Robot Half Marathon.
Scheduled to kick off on the morning of April 19 in Beijing Yizhuang, the event represents a milestone in the evolution of embodied AI. Organized by the Beijing Municipal Government and the China Media Group (CMG), the race will observe humanoid robots sharing the track with thousands of human runners, testing the boundaries of robotic stability, energy efficiency and autonomous navigation in a real-world environment Source: Toutiao.
For the team behind Walker Taishan, the stakes are as much about form as they are about finish lines. While some competitors are chasing raw speed, the Shandong-based team is aiming for a different kind of victory: the title of the “most human-like” finisher.
The Rigors of “Devil’s Training” in Jinan
The daily 21-kilometer loops at the Qilu Software Park serve as a high-stakes stress test. According to Zhou Tong, deputy general manager of Shandong Ubote Intelligent Robot Co., Ltd., the company has entered three teams consisting of seven to eight robots for the upcoming competition Source: Tencent News.
This “marathon rhythm” is designed to push every component of the robot to its breaking point before the actual race. For Walker Taishan, each run is a comprehensive examination of its hardware and software:
- Hardware Durability: The constant impact of 21 kilometers on real pavement tests the longevity of motors, joints, and structural components.
- Control Algorithms: The robot must maintain stability over long durations, managing energy consumption while resisting fatigue-like mechanical degradation.
- Environmental Adaptation: Running in a public software park forces the robot to interact with “temporary examiners”—unpredictable pedestrians, cyclists, and vehicles—testing its ability to identify and avoid obstacles in real-time.
The team is competing in two distinct categories: the autonomous navigation group and the remote control group. While both are designed for endurance, the autonomous group represents the cutting edge of the field, requiring the robot to map its environment and make split-second decisions without human intervention.
A Global Stage for Robotic Athletics
The 2026 Humanoid Robot Half Marathon is shaping up to be a significant international showcase. Recent full-process and full-element tests, completed between the evening of April 11 and the early hours of April 12, confirmed the participation of over 70 teams, including four international delegations Source: Toutiao.
The event is not just a novelty. it is a benchmark for the industry. By forcing robots to navigate a half-marathon course alongside humans, organizers are testing “embodied intelligence”—the ability of an AI to interact with the physical world. The challenges are immense: robots must handle varying road surfaces, manage battery life over several hours, and adhere to the rules of a crowded sporting event.
The Race for Speed: Breaking the 10m/s Barrier
While Walker Taishan focuses on “human-like” endurance, other competitors are pushing the limits of velocity. Unitree (宇树科技) has emerged as a frontrunner in the speed category. Their “Unitree H” model recently broke records by reaching speeds of 10 meters per second Source: NetEase.
The ambition in the field is rapidly escalating. Wang Xingxing, the founder of Unitree, has stated in public addresses that while humanoid robots have not yet reached the peak levels of human sprinters in 100-meter dashes, a breakthrough is expected by mid-year. Wang predicts that the running speed of humanoid robots, particularly those developed in China, will soon be “a bit faster than Bolt” Source: NetEase.
Technical Implications: Why the Half-Marathon Matters
To the casual observer, a robot running a race may seem like a spectacle. Though, from a sports analytics and engineering perspective, the 21.0975 km distance is a critical metric for several reasons.
1. Thermal Management: High-performance motors generate significant heat. Maintaining a steady pace for over 10 kilometers requires advanced cooling systems to prevent the electronics from throttling or failing.
2. Power Density: The “energy crisis” is the primary hurdle for humanoid robots. Balancing the weight of the batteries against the energy required to move a heavy mechanical frame over 21 kilometers is a complex optimization problem.
3. Dynamic Balance: Unlike a treadmill, a real-world marathon course has imperfections. The ability of a robot like Walker Taishan to recover from a stumble or adjust to a slope is the ultimate test of its balance algorithms.
| Focus Area | Walker Taishan (Ubote) | Unitree H (Unitree) |
|---|---|---|
| Primary Goal | Endurance & “Human-like” movement | Peak Velocity & Speed Records |
| Key Metric | Completion of 21.0975 km | m/s (Current record: 10m/s) |
| Training Method | Daily half-marathon distance runs | Sprinting and acceleration tests |
| Core Challenge | Structural durability & obstacle avoidance | High-speed stability & explosive power |
What Happens Next
As the countdown to April 19 begins, the focus shifts from the training grounds of Jinan to the streets of Beijing Yizhuang. The 2026 Humanoid Robot Half Marathon will provide the first definitive data on whether these machines can truly coexist and compete in human spaces.
The next confirmed checkpoint is the race start on Sunday, April 19, 2026. This event will determine not only which robot is the fastest or most enduring, but which architectural approach to humanoid movement is most viable for future real-world applications.
Do you think humanoid robots will eventually outpace human Olympic athletes, or is the “human-like” endurance of robots like Walker Taishan more valuable for the future of AI? Share your thoughts in the comments below.