Unitree's humanoid H1 robot reaches 10.1 m/s in the 100m sprint

Unitree's H1 robot can run at a maximum speed of 10.1 m/s. This is a peak value for humanoid robots.

listen Print view
Unitree H1 running.

(Image: Unitree Robotics / Screenshot)

4 min. read

The Chinese robotics company Unitree Robotics has achieved a speed of up to 10.1 m/s in a 100-meter sprint with its human-sized, humanoid robot H1. This is shown in a video that Unitree released on Sunday. However, Unitree does not state the total time the robot needed for the 100-meter distance.

Unitree conducted the sprint on a conventional athletics track. The humanoid H1 robot, approximately 1.78 m tall and weighing around 62 kg, reached a speed of 10.1 m/s, according to Unitree. However, this is not the average speed, but the maximum peak speed the robot reached on the track. Unitree also points out that a measurement error is possible.

The robot's achieved peak speed is thus slowly approaching the average speed of 10.44 m/s set by Jamaican sprinter Usain Bolt during his 100-meter world record in Berlin in 2009. At that time, the sprinter ran the distance in 9.58 seconds. Bolt also holds the world record for the 200-meter distance with 19.19 seconds, which he also set at the 12th IAAF World Championships in Berlin.

Empfohlener redaktioneller Inhalt

Mit Ihrer Zustimmung wird hier ein externes YouTube-Video (Google Ireland Limited) geladen.

Ich bin damit einverstanden, dass mir externe Inhalte angezeigt werden. Damit können personenbezogene Daten an Drittplattformen (Google Ireland Limited) übermittelt werden. Mehr dazu in unserer Datenschutzerklärung.

The H1 robot achieves high running speeds through its embodied intelligence combined with powerful components. The legs of the H1 robot have a total of five degrees of freedom. The leg actuators consist of gears and high-torque motors developed by Unitree itself. Unitree does not provide the exact specifications for the components used in the robot modified for the sprint. However, Unitree states the leg length ratios as 40 cm for the thigh and 40 cm for the lower leg. The robot draws the necessary energy for running from a 15 Ah battery with a capacity of 0.863 kWh.

The robot ran the 100-meter distance completely autonomously. Various perception systems, such as a depth camera and a 3D lidar system, assist it. However, during the 100-meter run, the robot deviated from a straight line, as the video shows. Therefore, due to its deviation from its running path, it would likely have been disqualified in an official race.

The official Guinness World Record for the fastest bipedal robot over the 100-meter distance is still held by the ostrich-like running robot Cassie with 24.73 seconds, set in 2022. Although the human-sized humanoid Tien Kung Ultra achieved a time of 21.5 seconds for 100 m at the World Humanoid Robot Games 2025 in China, this was only achieved through a time credit. The robot received the time credit because it ran completely autonomously compared to the other robots participating in the race. Its actual time was 26.87 seconds.

The robot Cassie comes from the robotics company Agility Robotics and was specifically modified for fast sprints by the Oregon State University Dynamic Robotic Laboratory. However, the robot is not a fully-fledged, human-sized humanoid robot like the H1. Cassie only has two legs and a small body without arms and a head. The robot has been fully optimized for fast running and can accelerate quickly from a standstill to the required high speeds.

Videos by heise

This is also the problem that developers of fast humanoid robots are struggling with: To break Usain Bolt's sprint record, such a robot must accelerate very quickly from a standstill and keep its body stable during the run to achieve an average speed of more than 10.44 m/s. Current human-sized humanoid robots are not yet capable of this. Nevertheless, the H1 robot's high running speed of 10.1 m/s represents a peak value for humanoid robots.

(olb)

Don't miss any news – follow us on Facebook, LinkedIn or Mastodon.

This article was originally published in German. It was translated with technical assistance and editorially reviewed before publication.