Humanoid robots’ 100-meter run highlights control challenges after the finish, The Nikkei reports

The Nikkei reported on Sept. 29, 2026, that a humanoid robot “crashed” into a wall after a 100-meter run. The article focused on why the robot lost control after running. The report said that while humanoid robots have demonstrated their ability to run, stopping and maintaining their course after finishing remain difficult. The episode highlights the stability of the final movement, not just the appearance of the time recorded.

For humanoid robots, stopping, avoiding obstacles and recovering their posture can reveal more about practical constraints than running itself. The report shows that demonstration speed alone cannot measure practical utility; end-stage control and safety also need to be assessed.

The article provides an example of why humanoid robots cannot be judged solely on whether they can run fast. A 100-meter run is an easy-to-understand measure, but the report focused on the robot making contact with a wall after finishing, putting the immaturity of end-stage control ahead of its speed. In other words, completing the run may work as a demonstration, but a full sequence of movements that includes stopping, slowing down and recovering posture remains a separate challenge. Our related story, FANUC’s U.S. shift in the AI era and parts supply chain reported by The Nikkei (2026-09-28), reported that companies are shifting toward the United States in the AI era while securing parts supply chains remains a key issue. Taken together, the reports suggest that robots are not complete simply because the hardware has been built; they approach practical use only when locomotion, sensors, control and supply chains work together. This also connects with government, DMG MORI and Komatsu to work together on data collection for physical AI, The Nikkei reports (2026-09-21), which described efforts to collect and standardize machine data. Reducing incidents such as post-run collisions or loss of momentum requires a foundation for continuously collecting real-world machine data and updating control systems. In the context of China’s humanoid industry, covered in The Nikkei reports on learning from China, a “humanoid powerhouse” (2026-09-18), mass production, pricing and a widening range of applications appear to be sources of competitive strength. A 100-meter run is an easy-to-understand way to showcase performance, but the topic may indicate that competition is shifting from demonstration speed to stable control, repeatability and safety. Three points warrant attention going forward. First, whether the wall contact was caused by the robot’s design, its control algorithm or the course conditions. Second, whether performance measures for stopping and fall avoidance will be published for similar running demonstrations. Third, how the distinction between showpiece competition and mobility for practical use will evolve.

For Japanese companies and research institutions working on humanoid robots, the issue is likely to be not only running performance but also how to manage stopping control and the accumulation of real-world machine data. In particular, without parts supply chains and a foundation for collecting factory data, it will be difficult to move from demonstrations to practical use.