"Science and Technology Innovation Board Daily" reported on September 25 (Editor Zheng Yuanfang) On the 24th local time, the official Tesla Robot account Tesla Optimus updated the "Embodied Intelligence Debut".
In the introduction, Tesla said that the humanoid robot Optimus (Optimus Prime) can classify items independently, and its neural network training is "completely end-to-end" - it can realize video signal input, Control signal output.
In the video, the Tesla robot demonstrated visual self-calibration, color sorting tasks, and the ability to maintain balance on one foot. It has many highlights in terms of perception, brain, movement and control:
In terms of perception, through visual perception and joint position encoders, Optimus can automatically calibrate the limbs and accurately locate the spatial position of the limbs.
In terms of the brain, relying on pure visual technology and completely locally deployed neural networks, it can quickly adapt to the environment and complete multiple tasks.
It sorts blue and green building blocks into trays of corresponding colors. Even if someone disrupts the blocks while grabbing the blocks, Optimus can immediately adjust and adapt to the new environment and continue sorting. Building Blocks. At the same time, it can also straighten overturned building blocks and perform new tasks such as "shuffle sorted building blocks".
In terms of motion control capabilities, Optimus can accurately grasp objects. When making movements, the robot's limbs, trunk, and finger movements are extremely flexible and close to humans. In addition, Optimus also performs multiple single-leg support sports stretching actions, and can maintain trunk balance while stretching.
It is worth noting that the input end of Tesla’s end-to-end model is a video signal, that is, a mixed input of image and audio signals, but this demonstration shows that its recognition part of the neural network algorithm only uses visual information.
CITIC Securities believes that it is feasible for the Tesla V12 autonomous driving algorithm system to be applied to both humanoid robots and cars, and will help accelerate the generalization ability of the recognition algorithm. As Tesla demonstrates the possibility of an end-to-end technology route for robots, the solution is similar to its FSD operating model. The successful experience of FSD and Dojo is expected to accelerate the iteration speed of robots and expand from single tasks to multi-tasks.
Looking at different links, (1)At the input end, analysts pointed out that the image recognition displayed by Tesla this time can be realized by 2D and 3D cameras, and the optional core is Due to cost control factors, domestic manufacturers of core components such as vision lenses and cameras already have the production capacity of such products, and product supporting solutions and product cost-effectiveness are worthy of attention. (2)
At the output end,Tesla Humanoid RobotThe output end is for signal and attitude control. Joints, sensors, and control systems are crucial to the stability of the output end, There are many options in this field, and different manufacturers are expected to have advantages in different price segments. (3)
On the algorithm side, the end-to-end framework can train directly from the original input end to the final output end, without manual feature engineering or intermediate stage processing. Although this video does not show its specific algorithm, based on its input and output information, recommends paying attention to the robot control algorithm based on the reinforcement learning framework. (4)
On the hardware side, the video shows the static motion control stability of the Tesla robot and the high accuracy of the IMU used to control balance, and the "dual coding" in the joint module The "encoder" does not have the disadvantage of strong resistance, and the encoder in the dexterous hand also has high accuracy. The importance of IMU, encoder and other sensing links is becoming increasingly apparent. In terms of specific targets, Zheshang Securities is optimistic about the general trend of industrialization of humanoid robots and focuses on the manufacturing leader of core components of humanoid robots.
Motor: MOONS Electric, pay attention to Inovance Technology, Jiangsu Leili, INCO, VEICHI Electric, etc.;
Screw: Hengli Hydraulic, Changsheng Bearing, Wuzhou New Year, pay attention to Best, Dingzhi Technology, Rifa Precision Machinery, Qinchuan Machine Tool, etc.;
Rewritten content: Reducer brands include Double Ring Transmission, Green Harmonic, Zhongli De, Shanghai Electromechanical, Hanyu Group, Fengli Intelligent, Han's Laser, Haozhi Electromechanical, etc.
Sensors: Huayi Technology, Keli Sensing, Hanwei Technology, Xindong Lianke, etc.;
Available controller brands include Huazhong CNC, Bozhong Seiko, Newstar, Eft, INVT, etc.
Lightweight: Seiko Technology, Zhongfu Shenying, Jilin Chemical Fiber, Jilin Carbon Valley and other companies are pursuing lightweight technology
Integration: Sanhua Intelligent Control, Top Group, etc.;
Complete robot machine: Provided by Eston, Boshi Co., Ltd., Xinsong Robot and other companies
Special robot: Jingpin special outfit.
Rewrite the content as follows:
The above is the detailed content of Tesla robot fully evolved! Perception, brain, and motor control capabilities are upgraded, and the end-to-end solution is showing initial results.. For more information, please follow other related articles on the PHP Chinese website!