Viral video shows humanoid almost breaking in half during Robot Olympics training

A humanoid robot crashed into a cushioned wall during a training sprint held in Beijing, sending sparks flying from its damaged frame. The robot was running at full speed when it appeared to lose control and failed to stop before hitting the barrier, with the impact nearly splitting its body in two. The incident occurred during preparations for the five-day World Humanoid Robot Games, dubbed the “Robot Olympics,” which is set to begin in Beijing on August 22. Almost 2,056 machines from 666 teams across 16 countries are set to tackle new challenges in the second edition of the World Humanoid Robot Games. Robot hits barrier A humanoid robot suffered a dramatic crash during a training session for the World Humanoid Robot Games in Beijing, with footage showing the machine slamming into a padded barrier and collapsing in half. Although it’s not clear which humanoid model it is, videos shared on social media show the bipedal robot sprinting down an indoor track at high speed. As it neared the end of the course, the robot appeared unable to slow down in time and crashed directly into the barrier. The impact caused the robot to bend sharply around its waist before folding forward and falling onto the track. Sparks could be seen as it struck the barrier. The footage was later widely circulated online and reportedly attracted more than 13 million views. A caption accompanying the video claimed that the robot had been moving too quickly to brake in time and had effectively “broken its waist.” The unusual crash prompted a wave of humorous reactions online. Some viewers questioned why the robot had not been programmed to recognize the barrier, while others speculated that its vision system may have struggled to distinguish the blue obstacle from the floor. According to experts, such failures highlight one of the main purposes of robot competitions: testing machines under demanding conditions and exposing weaknesses in perception, motion planning, balance and control. For developers, crashes can provide valuable data for improving robots before they are deployed in real-world environments. Humanoid events expand This year’s World Humanoid Robot Games features 30 competitive events and 21 scenario-based events, compared with 26 events and 487 matches at the inaugural games. The number of teams has risen 138 percent, while robot participation has quadrupled, according to China Daily. The US, Germany, Japan and other countries with established robotics industries will send teams to the competition. Brazil is participating with a five-member team of Robot World Cup champions. China accounts for the largest share of participants, with 157 companies, 641 teams and 1,975 robots taking part. Teams from 27 Chinese universities are also competing, representing participants from 30 provinces, autonomous regions, municipalities and special administrative regions. The scenario-based competitions have expanded considerably, growing from six categories last year to nine settings. They will test robots in environments including households, hotels, industrial facilities, emergency rescue, logistics, libraries, retail, office services and vehicle charging. Two new challenges are designed to assess practical robotic skills. The Dexterous Hand Challenge includes tasks such as tool assembly, powder weighing, block stacking, nail driving, bottle opening, unboxing, tweezer-based object handling and cable connections. The competition evaluates capabilities including vision, motion planning, force control and precision.The Integrated Five Challenge combines dexterous manipulation, stable object transfer and heavy-load lifting with a 100-meter (328-foot) run. Competition rules have also been tightened. Time limits for the 100-meter (328-foot), 400-meter (1,312-foot), and 1,500-meter (4,921-foot) races have been reduced to 1, 5, and 15 minutes, respectively. Most track and competitive events must be completed autonomously, with limited exceptions. Autonomy is also rewarded in scenario events, where fully autonomous performances receive a 1.0 weighting, while remotely controlled attempts are capped at 0.5.Get the latest in engineering, tech, space & science - delivered daily to your inbox.Jijo is an automotive and business journalist based in India. Armed with a BA in History (Honors) from St. Stephen's College, Delhi University, and a PG diploma in Journalism from the Indian Institute of Mass Communication, Delhi, he has worked for news agencies, national newspapers, and automotive magazines. In his spare time, he likes to go off-roading, engage in political discourse, travel, and teach languages.