- 7 weeks ago
- #technology
- #robots
- #innovation
- #future
- #ai
China is making sci-fi dreams come true with ultra-realistic robots—think synthetic skin, wild facial expressions, and even go-kart drifting! From household helpers to snake-like cable inspectors, the future is happening now. Want a peek at robots that learn, evolve, and even play soccer? Dive in! Subscribe for more mind-blowing tech and comment below with your favorite robot moment! #technology #robots #innovation #future #AI
👉 This channel was created in collaboration with https://www.youtube.com/@realidadeimpressionante
🔊🎧Music Info: Provided by the channel "MokkaMusic" and https://inaudio.org
0:01 - China's Rapid Robotics Advancements
1:00 - Biomimetic Humanoid Robots and Their Features
4:56 - Advances in Robotic Expression and Gestures
8:42 - Snake-Shaped Robots for Infrastructure Inspection
10:47 - Humanoid Robot Driving and Direct Perception Control
13:36 - Robots Learning Household Tasks in Real Homes
16:02 - XPeng’s Big Push for Humanoid Robots
17:52 - Humanoid Robots in International Competitions
👉 This channel was created in collaboration with https://www.youtube.com/@realidadeimpressionante
🔊🎧Music Info: Provided by the channel "MokkaMusic" and https://inaudio.org
0:01 - China's Rapid Robotics Advancements
1:00 - Biomimetic Humanoid Robots and Their Features
4:56 - Advances in Robotic Expression and Gestures
8:42 - Snake-Shaped Robots for Infrastructure Inspection
10:47 - Humanoid Robot Driving and Direct Perception Control
13:36 - Robots Learning Household Tasks in Real Homes
16:02 - XPeng’s Big Push for Humanoid Robots
17:52 - Humanoid Robots in International Competitions
Category
🤖
TechTranscript
00:01China seems to be turning science fiction into a stunning reality, right before our eyes.
00:07Ultra-realistic robots over there already have synthetic skin, facial expressions, and personalities controlled by artificial intelligence.
00:17Humanoid robots are already moving toward large-scale production and are starting to approach the market,
00:22while other machines are already working, patrolling infrastructure, and even entering real homes to learn from us, human beings.
00:31And the most impressive thing about all this is the speed at which everything is happening, my friends.
00:37Because while much of the entire planet is still discussing what the future of robots will be like, and ours
00:44too, of course,
00:45China seems determined to be living in that future before any of us.
00:49Let's take a quick look at the future that seems to have already arrived there, you know?
00:59Hello, impressed viewers. I'm Realidade Impressionante, and you're watching Realidade Impressionante.
01:08It's a fact that humanoid robots are getting better and better at walking, running, and performing physical tasks.
01:15But a Chinese company is working on another problem, you know?
01:19Making these machines also seem more human when interacting with us, human beings.
01:25During the 2026 World Robot Conference, held from August 19th to 23rd in Beijing,
01:31Noetic Robotics presented a new generation of biomimetic robots that are built around the same platform.
01:37Among the highlights were Xiao You You Wei, and also its male version, Xiao Shen,
01:43as well as characters like Xiao Sheng, Shen Yu, Xiao Ao, Shang Li, and Ye Lan.
01:49No, I'm not making this up. The names really are that strange, but I can comment on it here.
01:53Despite the visual differences, the idea behind these machines is relatively simple and basically the same.
02:00Instead of developing a new robotic system for each of these robots,
02:04Noetic uses a central architecture called X-Ridge-Rid.
02:08What they basically do is just modify, mainly, of course, the appearance, behavior, and personality
02:14using a different AI for each of these robots.
02:18The result is a kind of unique robotic platform, capable of taking on different identities
02:23without requiring major changes to its mechanical structure.
02:26And one of the most impressive parts is precisely the face.
02:30The head and face use a skin made of custom silicone and platinum.
02:34It is developed to reproduce physical characteristics, such as texture, elasticity,
02:39and even the small lines found on human skin.
02:42Underneath this artificial skin, there is a system with 32 active degrees of freedom.
02:47These mechanisms are what allow direct control of the facial regions to produce movements such as blinking,
02:52changes in social expressions, and also mouth synchronization during speech.
02:57But appearance is just one part of this system.
03:00The robots have also been given a multimodal perception architecture.
03:04Depending on the configuration, the set can include two RGB cameras positioned in the eyes,
03:09an additional camera in the body, depth sensors, lidar, ultrasonic sensors, and a four-microphone array.
03:16These sensors allow the system to visually track people,
03:20as well as interpret the environment and detect voice commands from other people.
03:25Noidix says it is already working to improve interaction in noisy environments.
03:29The acoustic system has been optimized to increase the accuracy of speech recognition
03:34and the sensitivity of the microphones.
03:36The response generation system has also been improved to maintain better conversational context
03:41and, of course, produce more natural responses.
03:44Even the audio system received special attention.
03:47The speaker, with a specially designed acoustic cavity,
03:50aims to produce a more engaging voice during interactions.
03:52But perhaps the most interesting aspect lies in the software,
03:56in the AI that controls this robot.
03:58The different characters can indeed share much of the same type of hardware,
04:02while individual characteristics are created through behavioral profiles and aesthetic changes.
04:08Senyu, for example, was introduced with a softer personality
04:12and an aesthetic inspired by a forest character.
04:16Zhao uses a much more striking appearance, associated with a desert character.
04:20Shuang Li, on the other hand, was given a more serene personality,
04:24while Ye Lan was introduced with a more mysterious identity.
04:27And their strategy follows an idea that is starting to appear more frequently in the Chinese robotics industry.
04:34Machines that keep evolving even after they leave their factories.
04:37Instead of delivering products with practically fixed capabilities,
04:41Chinese manufacturers want to create platforms capable of receiving new behaviors,
04:46functions, and even content over the time they spend with people.
04:49All of this in a way similar to what has already happened with smartphones and smart vehicles.
04:56And another Chinese startup, this one called AI Aheadform, based in Shanghai,
05:00also develops biomimetic robots.
05:03One of the main examples of this technology is the Origin F1,
05:06a robotic platform created for research,
05:09which involves embodied artificial intelligence and interaction between humans and machines.
05:14Instead of focusing development initially on the legs or the ability to carry heavy loads,
05:19the engineers started with the robot's face.
05:22Under an artificial skin made of silicone, which is flexible,
05:26dozens of small actuators control different regions of the face
05:29to reproduce movements of the eyes, mouth, cheeks, and other areas responsible for expressions.
05:34The Origin F1 uses up to 42 independent micro-actuators
05:38to reproduce the movements associated with the muscles of the human face.
05:42But of course, mechanically reproducing the human face is only part of the problem.
05:46And the evolution of the Origin F1 is starting to bring communication to other parts of the body, you see.
05:52On August 18th, 2026, a new demonstration showed the robot equipped with a hand with fully articulated fingers.
06:00In the video, the robot performs individual and also coordinated finger movements,
06:04as well as using its hands to make gestures during interaction.
06:07You can see that basically, the robot is able to wave and point.
06:11But this expands a platform that previously attracted attention mainly because of its biomimetic face.
06:18This change does add a new layer of communication that is non-verbal, using only the hands.
06:25That's because we humans don't communicate just using words and facial expressions.
06:29You can see how much I'm gesturing here, right?
06:31Hand movements, arm positions, body direction, and small gestures are also part of the way we convey information.
06:39So, Redforms is also trying to bring these elements to its machines.
06:43Other platforms developed by the company go even further in this direction.
06:48An example is Elf Chuan 2.0.
06:51In this system, the robotic structure has been expanded to also incorporate a more complete upper body, let's say,
06:57including a torso and articulated arms.
07:00This allows the combination of facial movements with gestures made by the arms
07:04and changes in posture during interaction with humans.
07:07The challenge then becomes coordinating all these systems in a unified way.
07:12The robot needs to identify the person, understand speech, generate a response,
07:17and simultaneously transform that response into voice, lip movements, gaze direction, facial expression, and body gestures.
07:25It seems simple, but believe me, it's still quite complicated to achieve, you know?
07:30These movements are part of an area known as embodied intelligence.
07:33That is, an AI in a robot's body.
07:36In it, that AI no longer exists only inside computers,
07:40and instead begins to physically perceive and interact with the environment around it.
07:44The AI in Headform Line includes different biomimetic platforms,
07:48some with very human features, which are very realistic, by the way,
07:52and others with more stylized designs, like the elf Wealth from Lord of the Rings over there, right?
07:56For now, these types of machines are still mainly research, development, and technology demonstration platforms.
08:04But the evolution of the E-Headform prototype shows how the company is gradually expanding this concept.
08:11First, their robots only had faces capable of producing micro-expressions.
08:15Later, they started combining these expressions with vision, language, and conversation.
08:20And now, the Origin F1 also starts using articulated hands and gestures during its interaction.
08:27Meanwhile, other platforms from the company are already advancing toward increasingly complete body structures.
08:32You see?
08:33So, tell me, in your opinion, what do you think their next step will be?
08:37I already know what you're thinking, but go ahead and tell me.
08:41And China isn't just about realistic robots, you know?
08:45Over there, they've also started using snake-shaped robots to perform a task that's difficult and pretty dangerous,
08:52inspecting power lines while they're still operating.
08:55The system was developed by Guangdong District Power Supply Boreal,
08:59and it's being used in Kuan Yin, the capital of Yunnan province.
09:02Instead of observing the cables from a distance, like drones do,
09:05these robots attach themselves directly to the power lines and move along them.
09:10The body is made up of several articulated segments,
09:13allowing the machine to get around obstacles it encounters along the way,
09:16including insulators and other components installed on the cables.
09:20At the front, there are cameras and sensors used to monitor the condition of the infrastructure.
09:25While moving, the robot can identify damaged cables,
09:29worn components, and abnormal temperature changes that may indicate the development of a failure.
09:34But a much more important feature lies in the way this machine receives its energy.
09:40The system uses a contactless energy harvesting technology,
09:43which allows it to draw electricity from the very line it is inspecting.
09:47With this, the robots can remain in operation for long periods,
09:51without being limited by the battery life of platforms like drones, for example.
09:55According to the officially released data,
09:57the robots have already traveled more than 130 kilometers of distribution lines
10:01and have achieved an inspection efficiency that is three times higher than traditional manual work.
10:06The technology also offers an alternative in places where drones face limitations.
10:11Some of the inspected lines are located near areas with flight restrictions,
10:15including regions close to airports, for example.
10:18In addition, drones used near high-voltage equipment may face electromagnetic interference,
10:24unfavorable weather conditions, and limitations related to flight endurance.
10:28Since the COBRA robots remain physically attached to the cables,
10:31these problems can be greatly reduced.
10:33And in this system, the robot doesn't need to leave the line to recharge, as we've seen, right?
10:38It patrols the very cables that also provide the energy needed to keep it running.
10:43This is really interesting, you know?
10:47And this other Chinese startup, also in robotics,
10:50has just come out of obscurity with a demonstration that's pretty common, okay?
10:54And here I remembered my dear friend, the CEO of Genesis,
10:57who has been sponsoring the channel for a long time now,
11:00João Galdino, who also has a humanoid robot.
11:03But in this case, the guys from China put a humanoid robot in a go-kart
11:07and taught the machine not only to drive and operate the go-kart,
11:10but also to perform drifts autonomously.
11:13The robot operates on its own.
11:15There's no remote control.
11:16The company is called Symbiosis Robotics,
11:19and according to the information released,
11:21it had been founded just about a month before this presentation.
11:25For the demonstration, the company used a ready-made robot from Unitree.
11:29This one is the G1,
11:30a commercial humanoid robot,
11:32about 1.27 meters tall and weighing around 35 kilograms.
11:36In the video, you can see that the robot needs to adapt its body to the limited space of the
11:41go-kart,
11:41hold the steering wheel, reach the pedals,
11:43and simultaneously coordinate its vision and movements to control the vehicle.
11:48According to Symbiosis, the operation is not performed by remote control,
11:51as I mentioned earlier.
11:53Behind the demonstration is the artificial intelligence model called
11:56Direct Perception Control, or DPC, according to them.
12:00The proposal, the idea, is to reduce the traditional separation
12:03between perception and movement control.
12:06In many of the robotic systems we see today,
12:09different modules are responsible for perceiving the environment,
12:13planning an action,
12:14and finally turning that decision into joint movements.
12:18The DPC uses a somewhat different approach.
12:21The model receives multimodal information,
12:23including vision, language,
12:25the robot's own body state,
12:27and tactile information,
12:28and then directly relates these to the movements of the robot's joints and hands.
12:32The company describes this architecture
12:34as a way to bring the brain responsible for perception
12:37closer to the cerebellum,
12:39which would be responsible for motor control.
12:41The goal is to allow the same model to learn skills
12:45from different types of bodies and situations.
12:48And driving a go-kart does offer a way to test several of these abilities simultaneously.
12:52The humanoid robot needs to visually follow the track ahead
12:56while it's turning the steering wheel with its hands
12:58and also, of course,
12:59controlling the accelerator and the brake with its feet.
13:02During a skid,
13:03these movements also need to be continuously adjusted.
13:06The robot also has to control its body
13:08while dealing with acceleration,
13:10braking,
13:11and changes in vehicle direction.
13:13Only those who've driven a go-kart understand what I mean.
13:17For robotic symbioses,
13:19integrated models of perception and control
13:21may represent a way to develop humanoid robots
13:24that would be able to learn increasingly complex physical tasks.
13:28And all of this without relying on a separate architecture
13:30for each stage of operation,
13:32as happens with most robots today.
13:36After running, besides racing go-karts,
13:38robots are also already running, dancing,
13:41and performing increasingly impressive demonstrations.
13:44Chinese humanoid robots are now starting to face a challenge
13:47that seems much simpler.
13:48Taking care of a house.
13:50Try telling that to a housewife and see what she says.
13:53Robotics companies in China are placing their machines
13:56in real home environments
13:57to learn tasks like picking up objects,
14:00organizing clothes,
14:01cleaning surfaces,
14:02and even arranging flowers.
14:04One of the companies involved in this project
14:05is X-Square Robotic,
14:07a Chinese startup backed by investors like Xiaomi and Alibaba.
14:11During a demonstration in Beijing,
14:13their robots performed different household tasks.
14:15But the company's goal goes far beyond showing off pre-programmed skills.
14:20X-Square has started bringing these humanoid robots into real homes.
14:23You know?
14:23In partnership with the Chinese service platform 58.com,
14:27the company created a service in which the robots work alongside human cleaning professionals
14:31in homes in Shenzhen.
14:33These robots have already entered and learned from over 100 homes in this program.
14:38During operations, they can pick up scattered objects,
14:40tidy surfaces,
14:41and perform other tasks,
14:43all while the human professionals are carrying out the more complex part of the cleaning service,
14:48for example.
14:48But there is a technical reason that is very important
14:51for putting these machines inside homes alongside humans for now.
14:55Homes are extremely difficult environments for robots.
14:58You know?
14:59In a factory, objects usually appear in predictable positions,
15:02and the machines can repeat the same movements thousands of times.
15:06Now, inside a home,
15:07where practically everything gets moved around all the time,
15:10a chair might be somewhere else,
15:12a piece of clothing might be folded,
15:14or it might be wrinkled,
15:15different objects can appear on the table,
15:18and each home has a completely different setup.
15:21Right?
15:21Even seemingly simple tasks like folding clothes
15:24require the robot to identify a deformable object,
15:28determine how to hold it,
15:29and adjust its movements while handling the fabric.
15:32In household tests,
15:34you can see that robots are still slow
15:35and can make mistakes during tasks that are relatively simple for us.
15:39In a residence in Beijing, for example,
15:41a robot from X-Square took about an hour
15:44just to fold a few pieces of clothing and organize the shoes.
15:48Even so, China is expanding its experiments,
15:51and other companies there are already having their robots
15:54train in regular homes as well, you see?
15:57In the next few years,
15:58we'll likely see robots in homes,
16:00at least in China, right?
16:02If you've been following me here for a while,
16:04you'll remember XPeng,
16:05that Chinese company that showed a humanoid robot
16:07with such a realistic walk
16:09that they had to cut the robot's clothes
16:11to prove it wasn't a human.
16:12You could see its skeleton right there, okay?
16:14The walk was really realistic.
16:17Well, now they're speeding up their plans
16:19to turn their humanoid robots
16:21into a new, large-scale business, you see.
16:24XPeng's robotics division,
16:26which is currently strong in electric vehicles,
16:28raised more than $900 million
16:30in its first round of funding,
16:32thus setting, according to the company,
16:34a new record for a single private round
16:36in the embodied intelligence sector in China.
16:39The company intends to invest
16:41a good part of that money mainly in iron,
16:43the humanoid robot I mentioned at the beginning.
16:45This machine was designed with proportions
16:47and movements that are similar to those of a human being
16:50and represents the company's attempt
16:52to transfer part of the technology
16:54developed for smart vehicles
16:55precisely to a robotic body.
16:58The company has also set a production goal.
17:01By the end of 2026,
17:03XPeng intends to reach a capacity
17:05of approximately 1,000 iron robots every month.
17:08The first robots will be deployed
17:10in the company's stores
17:11and industrial sites in China.
17:14Sales and deliveries to customers
17:16both in China and in other markets
17:18are expected to begin as early as 2027.
17:21But XPeng's strategy is not limited to factories.
17:24The company states that it intends
17:26to develop a general purpose robot,
17:28meaning the iron robot should gradually be prepared
17:31for different types of environments
17:32instead of remaining restricted
17:34to a single industrial task.
17:36The first commercial markets
17:37should include small and medium-sized businesses
17:40and commercial establishments.
17:42And then, of course,
17:43allow ordinary citizens to buy them.
17:45We've already noticed that China
17:47is at least sketching out the future,
17:51haven't we?
17:52Humanoid robots manufactured there
17:54took the field in Beijing
17:55with an important part of their intelligence
17:57developed on the other side of the world.
17:59Not there, here.
18:01A group of young Brazilian researchers
18:03took part in the 2026
18:05World Humanoid Robot Games
18:07using Chinese machines programmed with algorithms
18:10developed here in Brazil.
18:11The Brazilian team was made up of researchers
18:13from five universities
18:14and took part in the 5-on-5 soccer competition
18:17at an event that brought together
18:20666 teams from 16 countries.
18:22According to the source I found,
18:24which is Chinese, okay,
18:26among the members is Isabella Mora,
18:2821 years old,
18:29a researcher from Pernambuco
18:31who started working with humanoid robots
18:33about a year and a half ago.
18:34This happened when she joined the robotics team
18:36at the Federal University of Pernambuco.
18:38In July, she took part in RoboCup in South Korea
18:41before traveling to Beijing.
18:42But an important aspect of Brazil's participation
18:45is that the team didn't have to build
18:47their own humanoid robots.
18:49The robots used are manufactured
18:50by the Chinese company Booster Robotics,
18:53and this allowed the Brazilian researchers
18:54to focus their work mainly on the software,
18:57artificial intelligence,
18:58and the strategies used during the matches.
19:01During a match, for example,
19:02the robots need to locate the ball,
19:05identify their own position on the field,
19:07track other robot players,
19:09and decide how they are going to move.
19:11In addition, they also need to control their bodies
19:13in order to walk, maintain balance,
19:15and perform movements like kicking, for example,
19:18without losing stability.
19:19Most of these actions happen autonomously.
19:22In other words, the robot does this on its own.
19:24There's no one controlling it.
19:26For Flavio Taneo Dandel,
19:28the coach of the Brazilian team,
19:29putting the robots through progressively
19:31more complex challenges is necessary
19:33to develop machines capable of performing
19:35useful tasks outside the laboratories as well.
19:38The material I used for this video
19:39is based on the coverage published by Tim RUA
19:42on August 24th.
19:43And unfortunately, I didn't find anything
19:46on major news portals here in Brazil.
19:47If you know the people involved
19:49or are a participant, congratulations!
19:52If you'd like to share behind-the-scenes images
19:54or more info, my email is in the description.
19:57That would be very helpful.
19:59Thanks a lot.
20:00If you enjoyed this episode,
20:02you know what to do.
20:03Leave a like, make sure you're subscribed,
20:05and click on one of the two suggestions on your screen.
20:07And I'll see you in the next one, all right?
20:09I'm waiting for you, okay?
20:11Bye-bye.
20:12Click there and I'll see you.