- 2 months ago
- #technology
- #robots
- #innovation
- #ai
- #future
Robots marching in sync, border patrol bots, a rideable robot horse, and AI excavators—this episode has it all! Dive into the world’s wildest tech breakthroughs, from China’s shocking announcements to MIT’s futuristic controls. Don’t miss the coolest, craziest inventions! Subscribe for more jaw-dropping tech and comment below what blew your mind the most! #technology #robots #innovation #AI #future
👉 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 - Episode Introduction and Topics
0:46 - China's Humanoid Robots and Applications
3:39 - MIT's Natural Excavator Interface
5:59 - China's Dual-Method Tunneling Machine
8:17 - Robotic Horses and Quadrupeds
10:58 - Engine AI's T-800 Humanoid Robot in Industry and Combat
14:06 - Humanoid Robots for Security and Border Patrol
17:11 - AI-Driven Drone Defense System
👉 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 - Episode Introduction and Topics
0:46 - China's Humanoid Robots and Applications
3:39 - MIT's Natural Excavator Interface
5:59 - China's Dual-Method Tunneling Machine
8:17 - Robotic Horses and Quadrupeds
10:58 - Engine AI's T-800 Humanoid Robot in Industry and Combat
14:06 - Humanoid Robots for Security and Border Patrol
17:11 - AI-Driven Drone Defense System
Category
🤖
TechTranscript
00:00And in today's episode, among other things, you'll see that China shocked the world with the presentation of humanoid robots
00:08marching in perfect sync over there in Beijing.
00:11The United States is taking that idea of humanoid robots patrolling the border with Mexico seriously.
00:17A startup has created a robotic horse that is capable of transporting people and even other robots.
00:23A new MIT controller allows you to operate excavators using the natural movements of your arms.
00:29China has created a new tunneling machine that drills into the ground and uses explosives against solid rocks.
00:35Well, my friends, let's break down all this and a few other new tech stories.
00:45Hello, impressed viewers. I'm Carlos Salvez, and you're watching Incredible Reality.
00:5380 humanoid robots marching in perfect synchronization during the opening ceremony of the 2026 Humanoid Robot World Games held in
01:02Beijing, China.
01:03These machines in the video are the Booster T2 model developed by the Chinese company Booster Robotics and recently introduced
01:11as their new platform for research and development in embodied artificial intelligence.
01:16Well, the T2 is about 1.40 meters tall and weighs around 43 kilograms with 31 degrees of freedom.
01:24Twelve of them are concentrated in the hands, allowing for more precise movements during those manipulation tasks that we see
01:31robots doing out there.
01:32The robot also uses two sets of binocular cameras positioned on the head and in the waist area.
01:38This setup allows it to keep track of its surroundings, while another system maintains a closer view of the area
01:44where the hands are working.
01:46But one of the main changes compared to the previous generation is in processing capability.
01:51In the T2 Pro version, Booster uses the NVIDIA Jetson Tor platform, which can provide up to 2,070 teraflops
01:58of computational power to run artificial intelligence models directly on the robot.
02:03This local processing is very important for embodied artificial intelligence applications, where perception, decision-making, and movements need to happen
02:12continuously.
02:13In practice, the robot can observe the environment, interpret a situation, and turn that information into commands for its joints
02:20with less dependence on the cloud.
02:22Booster developed the T2 mainly as an open platform for researchers and companies.
02:26At the Beijing Humanoid Robot Olympics, this capability was also tested in a different activity than the walking demonstration at
02:34the opening.
02:34There, the Booster T2 was used as a platform in 7-on-7 Humanoid Soccer, where two teams compete in
02:41matches autonomously, meaning without human operators controlling each robot individually.
02:46In this scenario, the robots need to identify the ball, locate other players, navigate the field, and decide when to
02:53run, pass, or kick.
02:54I want to know when these guys turn into little cars. I want to see that.
02:58Therefore, soccer works as a test for problems that also appear outside of sports.
03:03A humanoid robot needs to perceive moving objects, avoid collisions, maintain balance, and modify its actions as the environment changes.
03:11These are very similar abilities to those needed for these machines to eventually work in factories, logistics centers, and other
03:19environments originally designed for us human beings.
03:21This presentation with 80 units of the Booster T2 robot showed another possibility.
03:27Operating a large number of humanoid robots simultaneously in the same space.
03:31Imagine an army there, huh?
03:33Something that, unfortunately, could become more and more common as these machines keep evolving, you know?
03:38And engineers at MIT and the United States have developed a new control interface for excavators that can drastically reduce
03:46the time needed to learn how to operate these machines.
03:48The operator, in a simple way, holds the mechanism and moves their own arm the same way they would when
03:55operating an excavator.
03:57By extending the arm, rotating it, or making a grabbing motion, the system interprets these gestures and turns them into
04:05the corresponding commands for the excavator.
04:07This proposal tries to solve one of the main difficulties faced by new operators.
04:12In conventional excavators, the movements made on the joysticks do not directly correspond to the movement the person wants to
04:18perform in the environment.
04:20The operator must learn which combinations control the rotation of the cab, boom, arm, and each machine component.
04:27Over time, experienced professionals develop a mental model that allows them to coordinate all these movements almost automatically.
04:34But acquiring this skill can take years of practice.
04:38Beginner volunteers and experienced operators were trained for one hour a day over seven days.
04:44When they used the conventional joystick controls, the beginners performed much worse than the professionals, which is obvious, right?
04:51Beginners performed much worse than the professionals.
04:54But with the World Space Interface, the name of this technology, the result was very different.
04:59The beginner operators were able to achieve performance very similar to the experts already in the first session.
05:05The researchers believe that the technology can also be used outside of simulators.
05:09A version of the system could be installed directly inside the cabin of an excavator, with the mechanical arm working
05:16almost like an amplified extension of the operator's arm.
05:19Another possibility would be to use the interface for teleoperation.
05:22In this case, a person could stay in a safe place while remotely controlling an excavator working in very dangerous
05:29areas, such as disaster zones, mines, or places at risk of collapse.
05:34Companies like Caterpillar, Hyundai, and Komatsu are also already working on simulators and remote operation technologies for these heavy machines.
05:42However, most of these systems still use joystick-based controls.
05:47The approach developed by MIT takes a slightly different approach, right?
05:52Making the machine directly interpret the movements that humans already make naturally.
05:59And China has presented what we call here the Tatu Zone, a new excavation machine designed to tackle one of
06:05the main problems encountered during tunnel construction.
06:08Unexpected changes in the geology of the terrain.
06:11This happens much more often than we think, you know?
06:14Called Shanglong, not Sheilong, okay?
06:17Or Lucky Dragon.
06:18The machine combines two techniques in a single machine that are normally used separately.
06:23It can excavate the ground mechanically, like a conventional tunnel boring machine.
06:27And when it encounters tougher rock formations, it uses drilling and explosives to break up the rock before continuing to
06:34move forward.
06:35According to its developers, this is the first machine in the world that uses two methods in this way.
06:41The equipment was developed by China Real Sense and Industry Group in collaboration with Xinhua University and came off the
06:48production line in Wuhan.
06:49The machine has a cutting head of approximately 4.5 meters in diameter and a hollow structure in its central
06:56region.
06:56Through this opening and other specific channels, drilling equipment can create deep holes directly in the rock formation in front
07:04of the machine.
07:05And these holes can then receive explosive charges.
07:08Before the detonation, the cutting head can partially retract, while the protection system absorbs the impacts and protects the sensors,
07:15mechanical components, and other precision parts that are installed behind it.
07:19The charges used are designed to produce fractures in the rock formation.
07:23After the explosion, the head returns to the working position and continues excavation, crushing and removing the previously fragmented rock.
07:30The system can also switch between different operating modes.
07:33Depending on the geology encountered, the machine can perform the first blast and then excavation.
07:39Or it can simply start excavation before using explosives, or work only with the conventional cutting head.
07:44It depends on the situation, of course.
07:46Other solutions have been incorporated to allow the equipment to adapt to the conditions encountered during the operation.
07:52Retractable side cutters allow the excavation profile to be modified,
07:56while a telescopic rear shield can adjust to different segments used in the tunnel lining.
08:01According to information released in China, tests indicated that the combination of the two techniques can increase excavation efficiency by
08:09approximately 30% compared to a conventional tunnel-boring machine in complex geological conditions.
08:17And Dashiehai developed a machine designed to function as a kind of mechanical horse, a robot horse.
08:24Called QGX-1, the robot attracted attention during the World Robot Conference 2026, held in Beijing,
08:31where visitors could literally climb onto the robot's back and take a short ride with this machine.
08:36The project combines four independent robotic legs with a structure equipped with a saddle and handlebars.
08:42In practice, the user rides the QGX-1 in the same way they would ride a horse or a motorcycle,
08:48for example.
08:48The difference is that, instead of wheels or muscles, movement depends on four mechanically and electronically controlled legs.
08:56And its dimensions are much larger than those found in most quadruped robots available today, these so-called robot dogs.
09:03According to the company, the QGX-1 can carry a load of up to 300 kilograms.
09:08This means it was not developed just to carry equipment,
09:11and it became quite clear that it has enough capacity to transport people.
09:16And in one of the most interesting demonstrations held in Beijing,
09:19even another robot, a humanoid robot, appeared riding on the robotic horse.
09:25That was really interesting to see.
09:27The machine was also designed to operate on surfaces other than perfectly flat ones, you know?
09:31Its four legs can adjust individually while the robot is moving,
09:35allowing it to cross uneven terrain and maintain stability while carrying the load on its own body.
09:41The QGX-1 reaches a maximum speed of about 10 kilometers per hour.
09:45Which is far, of course, from making it a replacement for motorcycles.
09:49But speed doesn't seem to be the main goal here.
09:52The advantage of a machine like this is precisely being able to reach places where conventional vehicles start to have
09:58difficulties.
09:58A wheeled vehicle depends on a relatively continuous surface,
10:02but a legged robot can choose where to place each of its limbs.
10:06This opens up possibilities for crossing rocks, slope terrain, trails, mud, and other surfaces where small wheels would have a
10:14lot of trouble.
10:15And the idea of building large quadruped machines big enough to carry humans is starting to show up in other
10:21projects as well, you know?
10:23Last year, we saw Kawasaki, for example, working on its own robotic horse concept called Corlew.
10:29It was originally introduced as a personal mobility machine capable of traversing difficult terrain.
10:35In this trade show scenario, at first a robotic horse might just seem like a curious demonstration, right?
10:41But a quadruped platform capable of carrying hundreds of kilos across terrain where traditional vehicles can't operate
10:47could find applications in agriculture, mining, equipment transport, rescue operations and logistics, warfare, and remote regions.
10:58The humanoid robot industry is starting to go in very different directions.
11:02And few companies show this as clearly as the Chinese company Engine AI.
11:07In just a few days, the company presented two completely different applications for its humanoid robot called T-800.
11:14In one of them, it's working inside a factory.
11:17In the other, the robots are fighting each other in a ring in the United States.
11:22The first demonstration took place at a Luxshare CT factory located in Shoshou, China.
11:28Engine AI released an official video showing several T-800 robots performing different industrial tasks inside a facility.
11:34According to the information presented in the video itself, the robots can receive work orders directly from the MES system
11:41used to manage production processes.
11:44After that, the vision system identifies the type.
11:49The robots then use both arms to pick up and position these objects.
11:54The demonstration also shows T-800 carrying a load with just its arms.
11:58And it does this by dynamically avoiding obstacles and performing full-body movements to position containers in a stable way.
12:05The choice of Luxshare for this demonstration is significant.
12:09The company is a leading Chinese electronics manufacturer and an Apple supplier.
12:13The presence of the T-800 inside one of these facilities shows the manufacturer's intention to take robots from demonstrations
12:20to industrial applications.
12:22But the video also brings up some interesting questions, right?
12:26The presentation is highly produced and edited, you can tell.
12:30And Engine AI keeps indicating throughout the footage that the scenes were recorded right there on site.
12:35And clearly this happens after previous demonstrations by the company raised doubts about the possible use of computer graphics.
12:43In the case I'm referring to here, Engine AI even released behind-the-scenes footage to show that the movements
12:49performed by T-800 were actually real.
12:52Remember the kick to their CEO?
12:54That's what happened there.
12:55Even so, in this new demonstration, the well-crafted editing makes it difficult to determine to what extent the robot
13:02is able to perform these tasks autonomously when it encounters situations that were not set up in advance.
13:07And the contrast becomes even greater when we look at what happened just a few days earlier.
13:13On August 15th, T-800 made its physical debut in the United States during REK-2.
13:19This is a humanoid robot combat event held in San Francisco.
13:22In this case, both T-800s were controlled by human pilots and were put face-to-face in matches broadcast
13:29over the internet.
13:30This tournament ended with the victory of a Japanese pilot named Hiroyuki Ozone.
13:35Wow, that's a hard name to say.
13:37These two applications are part of a strategy that Engine AI itself has been developing for the T-800.
13:43According to the company, the physical demands of the matches are used to test features such as structural resistance, dynamic
13:51balance, and joint torque.
13:52The T-800 has an initial announced price of approximately $25,000, while the company says it is working to
14:00reach a delivery capacity in the tens of thousands still this year.
14:05And the United States Department of Homeland Security, the DHS, has started talking to robotics companies about the possibility of
14:13using humanoid robots in security operations at the border between the United States and Mexico.
14:18Among the companies that have already presented their technology to the U.S. government is Fundation, the developer of the
14:24humanoid robot Phantom.
14:25They reported their idea is still in an early stage.
14:29The DHS itself states that, currently, it does not have any pilot project, agreement, or active contract with Fundation.
14:36But Sunkit Patak, the founder and CEO of the company, says that his team is already in talks not only
14:42with the DHS, but also with different robotic sectors of the U.S. armed forces, including the Army, Navy, and
14:49Air Force.
14:50The goal would be to use the humanoid robots mainly for surveillance and reconnaissance tasks.
14:54The Phantom was developed to walk autonomously over uneven terrain, map its surroundings, and distinguish humans from other objects it
15:03encounters along the way.
15:04This would allow these machines to be sent to regions where ground vehicles, fixed towers, or even drones have limitations.
15:12At the border, for example, the robot could patrol a certain area, identify movements, and then send an alert to
15:20a human agent.
15:21The decision about what to do would still be made by a person.
15:25That's what they say.
15:27According to Patak, one of the main advantages would be precisely to keep agents away from potentially dangerous areas.
15:33The humanoid robots could continue conducting surveillance for long periods, while human teams would only be called in when some
15:40relevant activity was actually detected.
15:43Another environment considered by the company is the tunnels found along the border.
15:48Since humanoid robots have a form of locomotion similar to humans, the Foundation believes they could indeed enter certain spaces
15:55originally built for people,
15:58and carry out reconnaissance there before human agents are sent to the location.
16:02But there's a pretty important problem here, you know?
16:05The current generation of the Phantom still hasn't been designed to continuously withstand the harsh conditions found in an operation
16:11like this.
16:25TAC believes that a small test could indeed be carried out quickly there, and that larger operations could be considered
16:32in the coming months, you see.
16:33There's an even more controversial possibility here.
16:37Although the company's current focus is surveillance and reconnaissance, the executive stated that Foundation would be willing to develop armed
16:44versions of these humanoid robots if the U.S. government requested that capability.
16:50Even so, he says that, at least in the near future, these robots wouldn't be responsible for automatically deciding when
16:57to use force.
16:58In other words, to take someone out, right?
17:01The machine would only detect a situation and send the information, and a human would still be responsible for the
17:07next decision.
17:11Drone swarms are creating an increasingly bigger problem for these modern air defense systems.
17:17Shooting down a single drone is relatively simple.
17:19The problem starts when dozens of them show up at the same time, and especially when each drone costs only
17:25a small fraction of the price of the missile used to destroy it.
17:29And now in the United States, Northrop Grumman has presented a rather different solution to this problem.
17:35Called Raid Hunter, the system combines artificial intelligence, a Bushmaster automatic cannon, and 50mm ammunition capable of correcting its own
17:44trajectory after being fired.
17:46Basically, it's a machine gun capable of piercing armored tanks in the hands of an A.I. deciding when to
17:53shoot.
17:53That could never go wrong, right?
17:55Let me explain that there's irony here. Some people might not get it.
17:58The system uses sensors connected to a defense network to monitor the airspace.
18:03A command and control system equipped with artificial intelligence processes this information and can identify and track multiple threats simultaneously.
18:12This includes small drones, in coordinated swarms, and even cruise missiles flying at low altitude.
18:18However, unlike a conventional cannon, the Raid Hunter doesn't just fire projectiles following a purely ballistic trajectory, okay?
18:25The ammunition used by the system has precision guidance.
18:29In practice, each projectile behaves almost like a small guided interceptor.
18:34If the drone is moving or tries to change its trajectory, the projectile can keep correcting its course as it
18:40approaches the cannon.
18:41For now, the Raid Hunter is still undergoing an accelerated testing campaign throughout 2026.
18:47Northrop Grumman states that they do intend to validate the system quickly so that it can be deployed in the
18:53very near future.
18:55If you like this episode, leave a like, check your subscription, and click one of the two suggestions on your
19:02screen.
19:02Just click, and I'll see you there.
19:05Thank you! Bye! See you soon! I'll be waiting. Click there!