ARTIFICIAL INTELLIGENCE IN SPACE APPLICATIONS

What's the role of Artificial Intelligence in Space applications ?

As we have seen in other articles, the general role of Artificial Intelligence is that of optimizing some functions to be able to perform them with the minimum error or adapt some behaviours of a machine to an unpredictable situation, namely, a situation that is difficult to be predicted during the design phase and that gives us some uncertainties.

The role of Artificial Intelligence in Space application is exactly the same: it is used to optimize some functions, performance, adapting a machine to unpredictable, uncertain conditions.

In this article, I will present a particular example of how Artificial Intelligence in Space applications is used. In particular, we will see how an automatic lander for comet/asteroids works, adapting itself to really uncertain conditions, difficult to predict during the design phase.

A brief introduction to asteroids landings

The history of landings on asteroids is not much extended. There have been only a few missions that allowed a lander to land on asteroids. 

The first one was Hayabusa from JAXA (Japanese Aerospace Exploration Agency). It was launched in 2003 and came back in 2010 with samples of the 25143 Itokawa asteroid. The second one was Rosetta from ESA (European Space Agency). It was launched in 2004 and reached the 67P/Churyumov-Gerasimenko asteroid in 2014. Unfortunately, a misfunctioning of a part of the intelligent system didn't work as expected and the lander overturned during the impact. The third one was Hayabusa 2 from JAXA. It was launched in 2014 and kept back samples of the 162173 Ryugu asteroid in 2020. The last one was OSIRIS-REx from NASA ( National Aeronautics and Space Administration). It was launched in 2016 and took some samples from 101955 Bennu asteroid. The probe is now coming back to the Earth.

All the systems used for these missions perfectly show the importance of Artificial Intelligence in Space applications. In particular, dampers used to dissipate the energy during the impact make use of Artificial Intelligence.

Let have an insightful explanation.

Artificial Intelligence in Space applications: damping systems of comet/asteroids landers

One of the biggest challenges of landing on asteroids is the impact phase with the soil. Since the mass of asteroids is really low (some tons), the gravity field generated by these celestial bodies is really low, almost negligible. For the same reason, the escape velocity is of some cm/s, meaning that if we have a minimum rebound during the impact, the lander will leave the asteroid forever

If we want to land on an asteroid, we have to dissipate all the energy during the impact, meaning that we have to design the best damper we have ever designed. But this not all.

The main problem during the design of a lander damper is that is impossible to know exactly which will be the conditions once we reach the asteroid. In particular, we can't perfectly know which the gravity will be and so which the impact velocity will be; also, the soil composition and the shape of the asteroid is known only approximately. All these factors give a lot of uncertainties during the design phase and for this reason, is very difficult to understand which performances our damper really needs. We can only have an unclear idea.

And here comes the importance of Artificial Intelligence in Space applications.

In fact, we can use Artificial Intelligence algorithms to making adapt our damper when it will approach the landing on the asteroid. Scanning the asteroid surface, measuring the velocity during landing and measuring forces and accelerations during the impact, we can make our lander "consciousness" of what is happening. Using these measures and Artificial Intelligence algorithms, we can actuate some components to change the properties of our damper and to make it generating the right damping force for that particular situation.

Let's see a real application.

A real example of Artificial Intelligence in Space applications: Piezoelectric active damper

A possible solution could be the usage of piezoelectric (PZT) actuators to apply a friction load to the damper piston. As we know, PZTs deform when they receive a voltage in input. If we install them on the internal wall of the damper cylinder, between the cylinder walls and the piston, we can power these piezoelectric actuators so that deforming, they push against the wall and the piston, generating friction and dissipating energy. By controlling the voltage, we can control the deformation of PZTs and so, we can control the friction. This finally means that we control the damping force!

So, we can give to PZTs different voltages input on the basis of which situation is occurring generating the friction that we need. What happens in the real situation is: 

• The sensors (accelerometers) measure forces and accelerations on the lander and these measures are sent to a computer; 

• The computer executes the algorithm and on the basis of the measures, the algorithm calculates the voltage that allows the PZTs to generate the friction that we need.  

• Finally, this voltage is given as input to the PZT actuator and the actuator deforms, pushing on the walls and on the piston and generating the friction force, namely, the damping that we need. 

All this process is automatic and continuous, we don't need to control anything from the Earth. The friction force continuously and automatically changes to achieve the performance we need.

This is the power of Artificial Intelligence in Space applications.

Conclusions on Artificial Intelligence in Space applications

In this article we have seen how Artificial Intelligence in Space applications works. Actually, as I said many time, there is nothing really intelligent, but everything works automatically through a perfect coordination of sensors, computer algorithm and actuators.

A real example of a PZT active damper for asteroid landers should have given an insightful view of which is the power of Artificial Intelligence in Space applications and should have better clarified, once again, what Artificial Intelligence is in general.

Of course, this is only one of the many examples of Artificial Intelligence in Space applications.

ScienceFull.