A humanoid robot named TianGong Ultra recently surpassed the 100-meter world record set by Olympic sprinter Usain Bolt by nearly one second, according to recent reports. The robot was developed by Jack Guo, chief engineer at X-Humanoid, who told Reuters that the achievement came after extensive development work, including long hours and the use of backup robots in case of damage. Earlier in the project, the robots faced significant challenges, with several experiencing severe falls and sustaining damage to legs and pelvises. One of the main difficulties the team faced was designing a reliable braking system. The robot can reach speeds of up to 17 meters per second while weighing 75 kilograms, but it lacks a way to stop abruptly. This has caused some robots to crash into the landing mat at the end of races. Future versions of the robot will be lighter, equipped with better braking systems, and designed to finish races in a safer posture. Guo explained that the focus is shifting from achieving high performance to ensuring reliability and practical use. He compared current humanoid robots to the early days of the automobile, saying that stability must be achieved before adding intelligence to perform complex tasks. Some practical applications are already emerging, such as a smaller model called Omni, which climbed 17 floors of the X-Humanoid headquarters during a test. Another robot in the series, Tianyi 2.0, was tested at Foton Cummins to move boxes weighing between 8 to 12 kilograms on shelves. Guo suggested that smaller humanoids could be used in service industries and light manufacturing, while the TianGong Ultra might be deployed for external inspections or emergency rescue operations. The braking challenge requires precise coordination between sensors, motor control, and trajectory planning to prevent falls. Without a strong deceleration strategy, the robot might slide or misplace a foot, leading to collisions or mechanical failure. Controlling the robot's movement involves algorithms that generate steps, adjust posture, and stabilize the robot in real time, even when unexpected disturbances occur. Unlike wheeled robots, humanoids are inherently unstable and must keep their center of mass within a range compatible with ground support while moving. Teams use physical models, fast control loops, and sometimes machine learning to handle unpredictable situations. The difference between a "high performance" demonstration robot and a "high reliability" robot for industry lies in the focus on measurable peak capabilities versus the ability to perform reliably and safely over thousands of hours. In industrial settings, managing safety margins and operational constraints is as important as hardware, with software playing a crucial role in making the transition from a one-time feat to a robust, usable system.