The Unitree robot's "hit-and-run" incident has become incredibly popular internationally! This footage is spreading rapidly across YouTube, Reddit, and X, with different angle versions circulating as viewers worldwide watch from all perspectives. Everyone is concerned that another human has been harmed by a robot. This already violates Asimov's robot principles: the first rule states that robots must not cause harm, or they will be immediately "destroyed."

But yesterday, after the Unitree robot collided with someone, it still won first place in the 1500-meter race, claiming the first gold medal at the inaugural Humanoid Robot Sports Competition. Is this... really acceptable?

Amid widespread discussion, this "accident" has been repeatedly analyzed, revealing that the blame may not entirely lie with the robot.

**Actually, It Was Humans Who Made the Error**

Yesterday, the Unitree robot achieved first place in the 1500-meter track event with a time of 6 minutes and 34 seconds (the human world record is 3 minutes and 26 seconds), marking the first gold medal of this humanoid robot sports competition and setting a world record. During the race, just as it was about to lap its opponents, it suddenly slowed down and began circling randomly around the track, displaying a rather show-off attitude.

Of course, when it comes to memorable moments, the widely discussed "hit-and-run" incident takes the spotlight. However, as sharp-eyed observers investigated further, the "culprit" behind this incident began to emerge—the cause of the Unitree robot's collision was very likely human error.

Just before the collision occurred, it was precisely the moment when two human operators were handing over the remote control. During the handover process, neither person noticed the person ahead and failed to issue a timely avoidance command.

Some netizens joked that this must be a tactic to defeat opponents. After all, high-level business warfare often employs the most straightforward methods.

But seriously, why do robot running competitions need remote control?

**Technical Reasons for Remote Control**

The specific reasons relate to the robot's stability, mainly due to dynamic balance issues. Running is a highly unstable form of movement, requiring robots to adjust their body posture in extremely short time frames. However, current sensors, algorithms, and actuators lack sufficient response speed and precision to handle various complex terrains stably. Human remote control can provide timely intervention when robots are about to lose balance.

For example, Unitree's another robot model G1 was used by a third party in a robot half-marathon in April this year, during which it fell during the race.

Another aspect is insufficient environmental perception. Although sensors can provide environmental information, during fast running, this perceptual data often has delays or insufficient precision, making it difficult for robots to make quick judgments like humans. Human remote control can compensate for robots' autonomous perception deficiencies.

The fact that the robot collision occurred just after briefly disconnecting from remote control perfectly illustrates this point.

It should be noted that robots are not completely incapable of autonomous movement without remote control. For instance, the runner-up in this 1500-meter race—the Tiangong Ultra robot developed by Beijing Humanoid Robot Innovation Center—ran without remote control.

This includes the "perpetrator" Unitree H1 (the same model that performed yangko dance at this year's Spring Festival Gala), which actually possesses certain autonomous movement capabilities without remote control.

H1 is Unitree's first general-purpose humanoid robot, born in August 2023. It stands 1.8 meters tall, weighs 47 kg, has M107 motors with 360 N·m torque, and each leg and arm has 5 and 4 degrees of freedom respectively. For perception, it's equipped with 3D LiDAR and depth cameras.

H1's movement speed is 3.3 meters per second, with potential performance exceeding 5 meters per second (18 km/h), while its average speed in this competition was 3.8 meters per second.

**Different Projects, Different Control Methods**

Returning to the question of robot sports, whether remote control is needed also relates to specific sport categories. Some projects currently aim to showcase robots' athletic abilities, such as free fighting, where humans also manipulate robots to punch and kick. This project tests whether robots' center of gravity is reasonable and whether they can remain standing after receiving heavy blows from opponents. This involves the design of hardware components like robot leg motors and joints.

Human control also enables more flexible and rapid offense and defense, allowing testing of robot hardware performance under intense confrontation.

However, the soccer competition is completed entirely autonomously by robots, as this project likely aims to test robots' strategic judgment and collaboration abilities. We can see that robots' soccer movements and flexibility are quite limited, with falls occurring from time to time. But when to kick, when to defend, judging who are teammates and who are opponents, and how to use tactics with teammates—these are all autonomous robot decisions.

Moreover, during matches, robots can regain form like humans. The algorithms remain the same, but second-half performance can be better than the first half.

It seems that humanoid robot sports competition projects mainly test their individual capabilities, whether "brain" or "limbs."

**Wang Xingxing's Response: Autonomous Running Next Time**

Don't be discouraged though—humanoid robot development is progressing much faster than we imagine. Unitree founder and CEO Wang Xingxing stated in yesterday's interview that robots will run autonomously in the next competition.

Yesterday at the sports venue, Wang Xingxing watched his company's robot compete. Regarding winning the gold medal, he said it was completely expected, and this wasn't H1's best performance, as it didn't reach the maximum speed of 5 m/s.

This good result was also a relief for him: "We've been criticized online so much in recent months that we couldn't stand it anymore."

He also revealed that Unitree robots will have more breakthroughs in the second half of this year and next year.

Regarding remote control use, Wang Xingxing responded that Unitree's H1 actually has the capability to operate without remote control, but due to different strategies and pursuit of speed, Unitree ultimately adopted remote control strategy on the field.

This approach certainly has drawbacks—it's "quite exhausting for colleagues." Operators running alongside experience severe physical exhaustion, so we saw on-site operators needing to relay to accompany robots through competitions.

In earlier interviews, Wang Xingxing also discussed the public's particular attention to robots "causing trouble." Currently, when robots walk or run normally, there's relatively little attention. But if robots have problems due to operations or other reasons, discussions explode exponentially.