What, robots have evolved to the point where they can build themselves? !
Researchers at the Massachusetts Institute of Technology (MIT) were so imaginative that they came up with this scene:
Robot No. 1 was busy picking out parts and assembling something. .
#After a while, the identical Robot No. 2 was born on the spot!
It's not over yet - Robot No. 2 just came into this world, and it immediately became as agile as No. 1. Then it "put on" two accessories for itself, oh no, accessories.
#Emm... It seems that the robot "matryoshka doll" has really come into reality now.
MIT researchers pointed out:
In addition to cloning itself, this kind of robot can also build larger robots in layers.
According to this idea, when building large buildings or large mechanical equipment in the future, it may be possible to construct small modules step by step like building blocks, without having to build a giant machine tool next to it.
This can greatly improve construction efficiency!
Relevant research papers have been published in the Nature sub-journal Communications Engineering.
It is not difficult to see from the previous animation that this kind of robot can clone itself so quickly, mainly Thanks to its piecemeal "parts," researchers call it Voxel.
#The word Voxel is the abbreviation of Volume Pixel. We can simply understand it as a module.
As for why the module grows like this, researchers say that the inspiration comes from the crystal lattice, which is the spatial structure in which atoms are regularly arranged in a crystal. The cuboctahedral structure has the advantages of low density and high stiffness, and is also easy to assemble and disassemble.
Previous modules were basically purely mechanical, and in addition to mechanical structures, these modules also have intelligent control systems.
The module is equipped with a battery, central processor, actuator, etc. in the middle, allowing it to move flexibly and navigate independently.
There are many electromagnets on the edge of the module, so that two adjacent modules can be firmly connected. Each set of face-to-face connections can transmit 8A current and 50N tensile force at 10V voltage.
But movement alone is not enough. The process of robots selecting parts and building new robots also involves grabbing actions.
So the research team also designed a "wrist" for the robot.
This "joint" is more flexible than other modules, making it easier for the robot to "pick up" parts and build them in layers.
At this point, there is still a question: How to grab the correct modules and spell out the desired new structure in order?
First, the robot's computing system will automatically discretize the input geometry into multiple ordered hierarchical building blocks through a compiler based on the given target shape.
Then, the system will also analyze the best assembly sequence to improve efficiency and prevent confusion in the "building block" process.
Next, the robot can clone itself in an orderly manner, and even build a larger robot.
However, the researchers also pointed out that during the actual operation, they found that the components were not flexible enough (the animation at the beginning is what it looks like in a simulation environment), and they are currently researching and developing more powerful and sensitive connectors.
The researcher comes from MIT's Center for Bit and Atomic Research and the U.S. Army Research Laboratory.
Amira Abdel-Rahman, the first author and corresponding author of the paper, is currently a doctoral candidate at MIT and holds a master's degree in technical design research from Harvard University.
Her main research direction is to use computers to analyze manufacturing processes and manufacturing systems.
In addition, the designer of the overall concept of the system is Professor Neil Gershenfeld, who is currently the director of the MIT Center for Bit and Atomic Research. He is also a member of the American Physical Society.
Currently, the MIT Center for Bits and Atoms Research is working to break the boundaries between computer science and classical physics, and explore how to better combine data and hardware.
Paper address: https://www.nature.com/articles/s44172-022-00034-3
Reference link: https://techcrunch.com/2022/11/22 /researchers-are-building-robots-that-can-build-themselves/
The above is the detailed content of The robot builds itself as easily as building blocks. For more information, please follow other related articles on the PHP Chinese website!