Tesla Optimus Zen 3 finally Here 5 HIDDEN SECRETS! SHOKING

Vídeo em Inglês, publicado pelo canal American Techie.

About this content: The summary, key points and topics below are generated automatically from the video transcript. They reflect the content, opinions, claims and predictions presented in the video and do not represent a position, validation or fact-check by IA Dots.

Video summary

This video discusses five key features of the Optimus Gen 3 humanoid robot, covering its hand design, split brain architecture, video-based learning capabilities, tactile fingertips, and onboard AI processing chip, alongside details about its battery and production timeline.

Key points

• Gen 3 moves 50 actuators into the forearm, using tendon-driven cables for lighter fingers, faster response, and reduced heat, achieving 22 degrees of freedom.
• The robot's brain is split into a reflex layer handling movement loops in about 50 milliseconds and a reasoning layer utilizing Grok for high-level instructions.
• Cortex 2.0 at Giga Texas converts human video footage through simulation into motor commands, allowing the robot to learn by watching tutorials or workers.
• Fingertips feature pressure-sensitive sensors that detect slippage and adjust grip within microseconds to handle delicate objects.
• The AI 5 chip allows vision, mapping, and obstacle avoidance to run locally on the robot without relying on the cloud, using cameras and the FSD stack for real-time 3D voxel mapping.

Topics

Optimus Gen 3Humanoid RobotActuatorsNeural NetsGrokAI 5 ChipBattery Pack

Video transcript

Transcript from available captions; it may contain errors.

unveil Optimus 3 in a few months. Um I think it's it's going to be quite surprising to people. It's an incredibly capable robot. We're replacing the SX line in Fremont with a million unit per million unit per year line of Optimus. Um now because it is a it's a completely new supply chain, there's really nothing from from the existing supply chain that that exist in in in Optimus. Everything is designed from physics first principles. The the normal S-curve of manufacturing ramp will be longer for Optimus than it is for products that have at least some portion of an existing supply chain. >> Everyone keeps saying Grok is running Optimus. Wrong. It's not even half the story. In this video, I'm breaking down five things inside Optimus Gen 3 that almost nobody has talked about. One, an engineering trick hidden in the hand. Two, a brain that's actually split into two parts. Three, a robot that learns by watching videos. Four, fingertips that can feel. And five, a chip that keeps the robot running even when the internet goes down. And at the end, the truth about the battery. Let's start with the hand because that's where Gen 2 struggled the most. In Gen 2, the motors were packed into the hand itself. Right where you need the most precision, you had the most weight and the most heat. And when heat builds up, the robot has to slow itself down. According to reports, Gen 3 flips that whole design. 50 actuators, 25 per arm, moved into the forearm with tendon-driven cables pulling the fingers. It's exactly how your own hand works. The muscles sit in your forearm and your fingers are just pulled by tendons. The result is lighter fingers, faster response, and far less heat. And the hand has 22 degrees of freedom. So, this isn't a gripper. It's a hand built to actually do work. Now, the most misunderstood part. Tesla has reportedly split the brain into two layers. The first is a reflex layer running on the same neural nets as FSD. Balance, steps, hand movement, all in a loop of roughly 50 milliseconds. The second is the reasoning layer and that's where Grok comes in. It only gives the big instructions, pick up the cup, put this part over there. Think about it. When you walk, do you think about every single step? No, your mind says go to the kitchen and your reflexes handle the rest. If Optimus had to wait on Grok just to stay balanced, it would already be on the floor. That's also why when you push it, it recovers instantly using foot sensors and internal IMUs. And the claim is that this kind of split is happening at this scale in a humanoid robot for the first time. Now, here's the part that sounds a little strange. Cortex 2.0, the cluster at Giga Texas, reportedly takes in human video footage and converts it through simulation into motor commands. That means the robot could learn to imitate a task just by watching a tutorial or a worker. If this fully works, every video ever recorded becomes potential training data. Back to the hand for a second. Every fingertip has pressure sensitive sensors. If an egg or glass starts to slip even slightly, the grip adjusts within microseconds. You can hold a glass with your eyes closed and never drop it. That was the hardest thing for a robot to learn. Number five, the AI 5 chip. Vision, mapping, obstacle avoidance, all of it runs inside the robot, not in the cloud. That means speed, privacy, and a robot that stays 100% autonomous even if the internet drops. No lidar, just cameras and the FSD stack building a real-time 3D voxel map of the room. Now, the battery truth. There's a roughly 2.3 kWh 52-V pack sitting in the torso. On light tasks, it can run up to 8 hours, but on heavy lifting, it drains much faster. Walking speed is 1.2 to 2.2 m/s, about the same as a person. So, enterprise rollout is moving forward from late 2026, but a robot in your home is still a way off. The hand, the split brain, learning from video, touch, and the chip, together, these five things are what set Optimus apart. Which one impressed you the most? Tell me in the comments. Like the video, subscribe, hit the bell icon, and I'll see you in the next one.

Assistir ao vídeo original no YouTube

Assuntos: Robótica e humanoides, Assistente pessoal, Robótica e humanoides. Publicado no YouTube em 2026-10-15.

Seleção automática por assunto e idioma. O vídeo pertence ao canal de origem.