This week, Figure AI, a robotics company invested by OpenAI, Microsoft, Bezos, and NVIDIA, announced that it has received nearly $700 million in financing and plans to develop a humanoid robot that can walk independently within the next year. .
And Tesla’s Optimus Prime has also repeatedly received good news.
No one doubts that this year will be the year when humanoid robots explode.
Sanctuary AI, a Canadian-based robotics company, recently released a new humanoid robot, Phoenix.
Officially claims that it can complete many tasks autonomously at the same speed as humans.
The world’s first robot that can autonomously complete tasks at human speed
Pheonix can easily Gently grab, move and elegantly place each object to its left and right sides.
It can independently identify the color of objects and place objects of different colors separately. Its operation is smooth and efficient, and its speed is close to human level.
Time Magazine named Phoenix the best invention of 2023, which shows that the mainstream mass media recognizes them.
Phoenix is controlled by the advanced and unique artificial intelligence system Carbon. It is 1.7 meters tall and weighs 155 pounds. It has the industry's leading robot hand. , comparable to the dexterity and fine operation of the human hand.
Carbon integrates modern artificial intelligence technology to convert natural language into real-world actions. It has interpretable and reviewable reasoning capabilities and is responsible for planning tasks and motion plans, making Phoenix like People think and act the same way to complete work tasks.
Its working load can reach 25 kilograms and can move at a speed of 5 kilometers per hour.
Some netizens made a table that counts the progress of the "evolution" of all robot company products towards humans. It can be seen that the top companies Already able to complete independent work tasks.
To complete precise work tasks, the main difficulty lies in hand flexibility and control technology.
According to Sanctuary AI’s demonstration, robots can now complete many simple tasks autonomously.
Check the test tube.
Classify screws and nuts.
Put the item into the bottle.
Scan the code to identify the item.
You can also complete the stacking game like humans.
Training method: autonomous learning from remote control data
Sanctuary AI trains Pheonix using remote control.
Staff use the VR system to manipulate the limbs, hands and fingers of the robot to complete various tasks.
Then, the data during the operation is placed in a simulated system to train the robot.
In order to realize the movements after training, and at the same time make the robot's movements fast, powerful, and precise, they chose a hydraulic system to complete the control of the robot.
According to CTO Gildert, only the hydraulic system allows the robot to "do some very dexterous things, such as threading needles and buttoning buttons. You have to ask yourself, which hand It can also lift a 50-pound suitcase and type on a keyboard at human speed."
In March last year, they put the robot into a supermarket in Vancouver. Let it complete 110 retail-related tasks. This includes picking and packing merchandise, labeling, labeling, folding and cleaning the store.
CEO: Geordie Rose
He is a serial entrepreneur in the technology industry.
Before founding Sanctuary AI in 2018, he founded 2 high-tech companies:
Kindred.ai: The world’s first company building reinforcement learning robotics for use in production environments.
D-Wave Systems Inc.: The world’s first quantum computing company, with contributions to Google, NASA, Lockheed Martin, Gap Inc. and several U.S. governments Institutions sell quantum computers.
# He holds a PhD in theoretical physics from the University of British Columbia.
He is also a combat athlete, a two-time Canadian National Wrestling Champion, and a 2010 NAGA Master White Belt Brazilian Jiu-Jitsu World Championship Dodo and winner of the No Way category.
CTO: Suzanne Gildert
She was with the CEO before, at Both D-Wave and Kindred hold important positions.
Prior to joining Sanctuary, Suzanne was the co-founder of Kindred AI. She oversees the design and engineering of the company's humanoid robots and is responsible for developing the cognitive architecture that enables the robots to understand themselves and their environment.
While at D-Wave, Suzanne developed MAXCAT – the world’s first game for quantum computers and developed the world’s first supervised classification to run on quantum computers machine and was the first to use a quantum computer to control the movement of a robot.
She graduated from the University of Birmingham, where her research direction was experimental quantum physics.
The two of them also regularly publish technology-related podcasts online, covering content ranging from corporate governance to specific technical details of the company. If you are interested in humanoid robots, don’t miss their exciting content.
Podcast address: https://sanctuary.ai/podcast/
The above is the detailed content of The first robot to autonomously complete human tasks appears, with five fingers that are flexible and fast, and large models support virtual space training. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
