Using Artificial Intelligence At NASA To Create Innovative Hardware

Jun, 2023 - by CMI

NASA is using AI to develop cutting-edge devices that will allow astronauts to explore uncharted territories and push the boundaries of human understanding. 

NASA uses artificial intelligence to produce mission hardware two-thirds lighter and more durable than man-made alternatives. The number of AI actions per minute is less than a thousand. Consequently, there are more refinement cycles and therefore more optimized designs can be produced faster. 

So far, the method has been used to build everything from the frame of the EXCITE telescope carried by NASA's balloon to the optical library used by the UV Imaging Spectrometer to store its optical components.

Like the ChatGPT chatbot or the DALL-E image generator, the system still relies on human intervention in the form of detailed commands specifying the requirements of the part, including the loads it must withstand and the forces it must withstand. must be submitted 

A generative The design program uses this data to build 30 to 40 iterations in less than an hour, each one improving on the previous one to create the ideal structure. The AI produces designs ten times faster than standard NASA procedure, which involves passing designs through designers, stress analysts who evaluate their performance, and mechanics who determine if they can be manufactured. 

It was found that AI-made components had ten times fewer stress concentrations while being two-thirds lighter. The company hopes the design approach will become standard procedure when creating structural components, electronics, and other subsystems in NASA instruments and spacecraft, as the agency manufactures thousands of parts. personalized each year for its many missions. 

This, in turn, will help reduce the time and costs associated with space exploration.

Because AI is so good at this stuff, it can dramatically reduce the cost of building these complex systems. The space station can hold six or seven people, but it costs $100 billion.