Humanoid Robot Poised to Break 100-Meter Dash Record: A Leap into the Future of Robotics

Recently, Sabastian Thor set a new world record with a sub-2 hour marathon. However, he is not the only athlete pushing boundaries. On April 19th, a humanoid robot from Chinese tech giant Honor shattered the human half marathon record. Additionally, Unitree’s robot is remarkably close to matching human speed in the 100-meter sprint. These advancements prompt two critical inquiries: How fast can humanoid robots run, and what purpose do they serve at such high speeds?

The inaugural Beijing E-Town Half Marathon and Humanoid Robot Half Marathon took place in 2025. This year’s edition witnessed a near fivefold increase in participating robot teams—over 100 teams fielded more than 300 humanoid robots. In 2025, the fastest half marathon time recorded for an autonomous robot was 2 hours and 40 minutes; this year, it was significantly reduced to just over 50 minutes source.

Meanwhile, robot manufacturer Unitree has announced that its bipedal H1 model reached an astonishing speed of 10.1 meters per second. For comparison, Usain Bolt’s record-breaking 100-meter dash requires an average speed of 10.44 m/s, indicating that the human record is clearly within reach.

Several factors contribute to the swift improvements in mobile robotics, according to Chef Petar Corum at Imperial College London. Reduced component prices, alongside advances in more powerful and efficient motors, have bolstered robot responsiveness and agility. Furthermore, faster computer chips utilize less power while enabling more sophisticated control algorithms. Enhanced communication and more compact, accurate sensors have also played a role in this evolution.

However, if speed is the primary objective, mimicking human design may not be the optimal approach. “Humans are not efficiently designed for running, as it’s not a survival necessity,” explains Benam Dadashzadeh from Bournemouth University, UK. In fact, research indicates that robots emulating an emu’s running style can be up to 300% more efficient than those designed with human features.

Dadashzadeh remains uncertain whether advances in mobile robotics will directly benefit the environments where humanoid robots are expected to operate. If speed is essential, “you can always simply equip it with wheels,” he suggests.

While the demand for mobile robots may not be evident, competitions serve as excellent testing grounds. Kormschev notes, “These events act as stress tests for hardware, requiring high torque over extended periods, which can lead to overheating.” He compares it to car manufacturers participating in rigorous rally competitions, which demonstrate their ability to produce durable products. Both Unitree and Honor declined to comment to New Scientist on their motivations.

However, such competitions can lead to designs that may not be well-suited for practical applications. Kormschev points out that robots showcased in running events often lack functioning limbs or even heads, with large hip joints optimized purely for straight-line speed. “If lateral movement is required, these robots will struggle, as their design favors forward motion at the expense of versatility,” he explains.

Just because humanoid robots become more capable and affordable doesn’t imply they will lack utility. Human-like robots hold several advantages in spaces designed for human interaction, including the ability to operate door handles, ascend stairs, navigate furniture, and use tools.

So, how fast can humanoid robots ultimately go? Dadashzadeh suspects that the upper limits for human-like robots are likely already established. He predicts that while they may surpass human records, the difference will be slight. “They’ll be close, but robots will likely be just marginally faster,” he states.