A marathon stripped the bottoms of Chinese robots? It's a bit humorous

A few months ago, I was still changing handkerchiefs at the Spring Festival Gala, and now the humanoid robots have already run marathons on the road, and this weekend is really magical.

It may be cool to imagine, the school sports meeting is held in Cybertron, and the thunderbolt is chasing Optimus Prime; In fact, if you want to watch the live running on the Internet, you will feel deeply abstract:

Some of them lost their heads halfway through the race, some fell into baby-like sleep at the beginning, and some looked like people and wanted to lie down on the ground as soon as they got on the track.

There is even a winged one, as soon as the machine is turned on, it will be thrown away by the pilot, and it really doesn't look like a human form, and it's not a human thing.

Aside from these abstract players, there are also many players who focus on the game.

Like Tiangong Ultra from Beijing, it perfectly gives full play to the advantage of one meter and eight long legs; The little urchin N2 with loose power has short legs and can only run wildly;

There is also Xing No. 2 from Shanghai, whose family won the world record of robot walking ten years ago, and the sect has a very deep heritage.

In the end, Tiangong Ultra won the championship, and sure enough, you have to look at the dark skin when running, and the sect's heritage is difficult to compete with racial talent after all.

Although the onlookers on the side of the road all thought it was quite joyful, as soon as they saw the live broadcast screen, some people immediately began to be surprised and questioned, some filters were broken, some were finally disenchanted, and many of them were deceived and had to precipitate.

For example, some netizens post videos, your robot has a remote control, can this still be called a robot? That's not how you draw cakes.

There are also some netizens and foreign media who say that they are still far ahead of the level of this move? I know that Yushu is CG deceiving people, and even said that this is a Chinese robot bottom, and it is really a slap in the face, a chicken feather.

Should I say it or not, if you look at the performance, this is indeed a bit of a contrast with everyone's previous fantasy of robots.

But if you're disappointed with the bots and turn black, then I think it's too early to draw conclusions.

Let's put it this way, first of all, there are many teams participating in this competition, but because the threshold for participation is not very high, many of them are actually just here to bask their faces, and some are even personal enthusiasts.

If you want to take these rollover cases to prove the garbage of the entire humanoid robot industry, it is completely logically wrong, and it is a hasty generalization, which is a partial generalization.

For example, although this competition is huge, a leader like Yuki has not come at all, and people are busy engaging in robot fighting.

As for the cases of Unitree robots overturning, they are actually because these participating teams bought Unitree robots by themselves, but did not use Unitree algorithms, but used their own.

In addition, the first heavenly work in the end happened to be the robot produced by the organizer itself. In this competition, there are only a few companies that are really making product-level robots and adapting algorithms by themselves.

In other words, this competition is for Tiangong and other top few to show their muscles, staring at the overturning cases of other teams to spray Yushu, which is purely crying the wrong grave.

Yushu didn't come at all, so why don't you leave Kuroko here?

In addition to logical problems, these netizens actually have a bit of a blind spot in their cognition of humanoid robots.

Many netizens say that the remote control is not a robot, and there are even toy remote control cars as an analogy, but this analogy is not appropriate.

Although running is the most basic function of a humanoid robot, it is much more difficult to implement than a toy car.

The main humanoid robot has too many degrees of freedom, compared to the fact that the vehicle only needs to drive four wheels to rotate, and the robot has 14 degrees of freedom to coordinate its lower limbs alone, and it is more complicated to count the arm and torso movements, and it is even more difficult to control these body parts.

If you want the wheels to turn and you can give electricity, but if you want to let dozens of degrees of freedom listen to you, this kind of pure line control will not work, and even the traditional Lagrangian equations and Newton-Euler equations are difficult to do, and you have to rely on more complex dynamics modeling and algorithms.

Even if you can finally control every joint of the robot to rotate as you want it to, balance is a big problem. As long as the car has four wheels, he is stable, but the robot has two feet, and the center of gravity is high, so it would be good to be able to stand.

In order for him to run, he has to alternate between one-foot support and take-off, and he has to maintain balance throughout the whole process without falling, which not only uses the center of mass (CoM) trajectory prediction, but also considers the zero moment point (ZMP) dynamic adjustment to achieve stability, and also needs to study how to achieve instantaneous feedback in unstructured terrain (such as slopes, gravel, mud, grass), ground friction coefficients, obstacle collisions and other external disturbances, and complete the calculation of joint parameters (such as position, velocity, acceleration) in milliseconds. Correct model predictions in real time.

In a word, this thing is not at the same level of difficulty as making a toy car or a toy model robot. Just like you can rub your hands in the monkeys, it doesn't mean that you can send a living person to the moon and back alive.

Therefore, it is irrational to just see the surface of things and deny them directly. The control of humanoid robots with a high degree of freedom has always been the pinnacle challenge in the field of industry and algorithms, and it is not bad to be able to run 21 kilometers.

After all, if you think about it, in the carbon-based biological world where tens of millions of muscle fibers are mobilized by the central nervous system to achieve precise control, there are only a few animals that can run on two legs.

Then there is a bad friend who is going to ask.,You're a pure blower.,You can blow it even if you're so abstract.。

Not really. Although it is true that some of the robots in this game are running very abstractly, some are simply walking, and some have their heads running away, there are also many people who see the expectation from this game and see the vitality of all things competing.

Because compared with these failed cases, Tiangong Ultra, Little Naughty N2, and Walker 2, which can complete the race and even run well, also show China's real progress in the core technology of robots, such as motor power density, motion control algorithms, and dynamic balance.

To run these 21 kilometers, it means that the robot has to complete about 250,000 precision joint movements, which is a great test for the robot's endurance, components, and algorithms.

According to the Beijing News, Xiong Youjun, general manager of the Beijing Humanoid Robot Innovation Center, said: "This is not only a simple sports competition, but also a test of technological breakthroughs and industrial development." ”

Tang Jian, the company's chief technology officer, also believes that the main purpose of this competition is to test the stability and reliability of the robot, which is equivalent to the experience service of the robot for 7× 24 hours in the future.

Team No. 2, who also won a medal, said that the competition was not to win the ranking, but most importantly to test their skills. For example, can you run from beginning to end without changing the battery. Because endurance is a problem that has plagued the field of robotics for decades, it is much more significant to solve this problem than running a marathon to get a ranking.

In fact, the beginning of every technological revolution is accompanied by various imperfect and even ridiculous attempts.

For example, in 1894, France hosted the world's first competitive motor racing competition, and the 21 cars involved in the competition had steam engines and internal combustion engines, and the race scene was quite abstract.

There were various car accidents on dozens of kilometers of road, some cars broke their axles as soon as they started, some cars ran half of the wheels and fell off, some steam engines exploded and sent the driver to the hospital, many pedestrians were hit on the scene and three dogs were killed, and the first place to finish the race was actually a steam tractor.

For example, at the beginning of the 20th century, there were many aircraft competitions in the United States, monoplanes, biplanes, triplanes, and even pentatoms and all kinds of creative aircraft designs participated, and there were many abstract things that happened.

In the 1910s, the "Demoiselle" aircraft designed by Santos Dumont was nicknamed "capsize" by French journalists because of its rear center of gravity and easy to roll forward when landing. In the 1920s, the stunt group "13 Black Cats" caused accidents with clear prices in order to attract audiences.

During the 1925 Pulitzer Games, a 16-year-old homeless man tried to pick up a plane to get to Las Vegas, and he was blown naked in the 90 mph wind, leaving only "sleeves and collars";

In the 1927 Dale race, only 2 of the 15 teams crashed into the sea due to navigation errors, lack of fuel or mechanical problems;

Lowell Bayles, the 1931 Thompson Cup winner, crashed during a record-breaking flight due to a detached fuel cap or a body tremor that ran out of control; From 1930 to 1932, the Granville Gee Bee series was known for its speed, but several award-winning pilots lost their lives one after another.

It would be easy to conclude that cars are inferior to horse-drawn carriages, internal combustion engines to steam engines, and airplanes to be dangerous jugglers, but this is undoubtedly absurd in the present view.

Should I say it or not, from the national explosion of the Spring Festival Gala robot, the big win and the special win, the pursuit of investors in the past few months, to the recent doubts of some investors, and then to the decline of many netizens after this robot marathon, the change of mentality before and after is a bit irrational.

But I think we might as well watch the excitement with a little more fun mentality, and a little less extreme emotion of quick victory or surrender.

Criticism is necessary, and it can spur the industry not to just paint cakes, but to come up with the real thing. But to be overly pessimistic, because it rises to the level of picking bottoms and chicken feathers in one game, it is a bit of an overreaction.

In short, laughing and squirting, we must also see that those engineers who are working hard are trying to turn "abstraction" into reality. It's better to give the skills some time than to win or sing without brains, and maybe the next time, they can run more steadily and faster.

Rome was not built in a day, and whether robots can go from an abstract marathon to thousands of households is not something that can be finalized in a race.