If every part in an assembly had intelligence, products could build themselves like proteins

Skylar Tibbits from MIT's Self-Assembly Lab thinks that manufacturing is moving too slowly.

https://www.solveforx.com/moonshots/can-we-make-things-that-make-themselves

In this TED talk he points out the long build times of skyscrapers and space shuttles, and then contrasts the nanosecond timeframes used by complex proteins and DNA replication.

Natural systems are more efficient and more complex than human construction. Nature doesn't make mistakes, uses energy wisely and can repair itself.

Tibbits presents a plan that begins with decoding the assembly order of things and ends with correcting internal errors.

This talk is full of projects and demonstrations from the MIT Self-Assembly Lab.

The MacroBot and DeciBot are large scale self assembling robots, embedded with sensors and electromechanical devices. Each element takes information in, checks its progress, and transmits the assembly information to the next part.

http://sjet.us/PROJECTS/MIT_DECIBOT/dsc_5468_pshop_small.jpg

The LogicMatter consists of several tetrahedrons. A digital logic gate can be imprinted into the part to determine how a structure will be assembled. Through inputs from the user and from the previous piece, the components transmit an output to the next piece, all in three dimensional space.

http://www.suckerpunchdaily.com/wp-content/uploads/2011/04/logic_lg.jpg

Finally Skylar demonstrates biased chains. He takes a chain of components, all identical and biased, and literally shakes them until they assemble into different predetermined shapes. Each element in the chain is told if it should turn left or right based on the location of the piece before it.


http://www.sjet.us/PROJECTS/MIT_BIASED_CHAINS/ShakeNFold_small.jpg

The talk is incredibly inspirational and full of possibility. Tibbits is a very engaging speaker and his self-assembly and 4d printing ideas have recently been getting wider exposure.