How AI could revolutionise space travel


Artificial intelligence is offering improvements to all kinds of industries (1). And space exploration is no exception. In addition, AI is developing so rapidly that by 2024 – which is when Elon Musk plans to send humans to Mars – AI will be capable of efficiently carrying out many tasks that are crucial to space travel (2). A recent technological innovation from Japan emphasises how computer intelligence could deliver wide-ranging benefits to future space missions.


Meet the ‘Int-Ball’

The Japanese Aerospace Exploration Agency (JAXA) has sent an autonomous robot to the International Space Station (ISS) (3). The robot is called ‘Int-Ball’: a spherical camera drone that can send pictures and videos back to Earth to assist with repairs and other tasks.

This drone can be controlled remotely or autonomously and can easily move around the microgravity environment of the ISS. Int-Ball records three-dimensional target markings, and by using this information it can determine its position and orientation. The drone is able to take photos from any location and at every angle.

Utilising Int-Ball means that crew members have one less task to deal with. Normally, astronauts have to handle a camera to send this vital information back to Earth. This currently takes up 10% of their time. Freeing up this valuable time is one way in which robots will make space missions easier and therefore more likely to run smoothly.

JAXA says that they are working on improving Int-Ball’s capabilities and functionality. Indeed, it will be interesting to see what robots will be capable of doing by 2024.


Will robots replace astronauts?

UK AI firm DeepMind has taught computers to reason about the world; more specifically, to understand how different objects are related to each other. This is called relational reasoning and it involves figuring out, for example, whether something is to the left or bigger than something else. Being able to transfer these abstract relations from one domain to another is a fundamental part of human intelligence.

A team of researchers were able to train the AI system to analyse images of three-dimensional shapes of different sizes and colours. Then they asked it questions, such as ‘What size is the cylinder that is left of the brown metal thing that is left of the big sphere?’ And the system would answer correctly 95.5% of the time, which is slightly better than humans.

However, Adam Santoro, who led the study, said practical applications of the system are a long way off. AI still can’t really make sense of the messiness of the world like we can. Human reasoning and problem solving, in general, is very broad, and it is difficult for AI systems to make many of those fast, intuitive judgements that an astronaut will have to make. Moreover, when it comes to relational reasoning specifically, this is a type of reasoning that involves so much more than just understanding differences in size, colour and shape.

There is a similar story in terms of those other human traits that every successful astronaut needs: perception, learning, natural language processing, motion and manipulation, social intelligence and general intelligence. With respect to each of these abilities, great strides are being made in the field of AI technology. But human-like capacities are still very much lacking. Computers can beat us at certain games (4), but AI cannot compete with astronauts (not yet, at least). While AI may not replace astronauts by 2024, researchers believe that in 45 years there is a 50% chance of AI outperforming humans in all tasks (5). So by this time, it may be unsurprising to see intelligent robots doing what astronauts do best.

The benefits of replacing astronauts with AI include huge increases in efficiency (robots don’t need to sleep) and all kinds of savings (AI is programmed, whereas people have to be trained; and robots don’t need food or water in order to survive). Also, if astronauts were completely replaced by machines, then you’d avoid adverse psychological effects (6) (assuming the machines are intelligent but not sentient) and various health complications. Thus, AI may play a pivotal role in improving the success of future space missions.


About the author: Sam Woolfe @samwoolfe

Sam is a writer who is especially interested in space exploration, sustainability, animal agriculture, nutrition, wellbeing and smart drugs.





Please enter your comment!
Please enter your name here