Aerospace Agency NASA somewhere in the 2030s wants to send a manned mission to Mars. To reach the red planet safely, the Agency will need the most advanced and reliable space technology. One of these technologies can be artificial intelligence, which will be used to control almost all the systems of the spacecraft, and in the future and the space dwelling, which people will equip on our planetary neighbor. Development of AI of this level is already underway and, interestingly, the prototype for it was perhaps the most infamous machine from science fiction – computer HAL 9000.
Robotics and expert in the field of AI, Pete Bonasso of the company TRACLabs (Houston, USA) said that he developed a new system prototype CASE (cognitive architecture for space agents) in technical terms is completely similar to the computer HAL minus such is clearly unnecessary for machine hell, sociopathy, paranoia and izmenitekst.
If we do not take into account these psychological shortcomings, the advanced computing power and abilities of the cult character from science fiction of the late 60s made a very vivid impression on Bonasso half a century ago.
“The first time I saw “Space Odyssey 2001″ by Stanley Kubrick in 1968, when I was in my senior year Academy West Point”, — writes Bonasso in an article published by the journal Science Robotics.
At that time, students had access to only one computer of the Academy. Not for any whim of the teachers or Director, just for the whole University had only one computer. Such machines were rare at the time. This computer was a General Electric 225 station equipped with only 125 KB of RAM. Despite the limitations of the machine, Bonasso quickly figured out how to program it to play virtual Billiards. But when a young student saw the possibilities of the Hal computer, it was a real revelation for him.
A few decades later, already a specialist in the field of AI Bonasso developed what he saw in the film of the 68th year.
The AI prototype created by Bonasso was able to control the computer-simulated environment of the simulated space station for just four hours, but the results are already encouraging: the program did not kill any virtual astronaut during the simulation.
“Our colleagues, as well as representatives of NASA, with whom we cooperate, do not worry that HAL can get out of control. Why? Because the system cannot get out of control. She does only what has been programmed”, — says Bonasso.
The main function of the CASE system in the future will be to manage all the activity and technological operations of the space colony, so that it works like a clock. The system architecture consists of three layers. The first deals with the management of hardware, for example, life support systems, electrical network, planetary Rovers and, Yes, theoretically, the doors of the airlock compartments.
The second layer is intended for software management, on the basis of which the hardware infrastructure will work. In addition, this cluster will be responsible for compliance with the daily routine tasks, for example, to conduct a daily health check of all systems of the station, as well as to prevent and solve potential emergency situations (gas leakage, fire, broken generators, approaching dust storms, and so on).
Presumably the third layer of the CASE architecture will deal with the problem of the alien monolith, if it suddenly appears next to the colonists, but the published article says nothing about it.
In addition to the multi-layered architecture, CASE will have an ontological system that gives AI the ability to reason and analyze information that will come to it through machine-human interaction interfaces (for example, visual displays and digital dialog boxes that will respond to speech).
Most of the above functions prototype, of course, is currently able to perform only in a virtual environment, but Bonasso and his colleagues from TRACLabs, a company that collaborates on the development of new high-tech systems, for example, with the same NASA and other government agencies, hope to soon move the test system from the virtual world to the real.
“We are working with experiments aimed at modeling systems. As part of these tasks, people get together and simulate the environment and life on distant planets or the moon. We hope that step by step, but we will be able to implement the CASE in reality and assess its importance for future space expeditions,” the article says.
Looking ahead, if such real systems prove their worth in real tests and eventually will be used in the framework of colonial missions on the moon and Mars, according to Bonasso, they can really simplify the development of deep space.
Again, the scientist is sure to worry about the risks of a riot, as shown in Kubrick’s film, not worth it. The capabilities of such systems will be limited only by the set of functions that have been programmed into them.
“The abilities of the same CASE, even in a virtual environment, are very limited and focused on the events that will take place on the planetary station. This AI will be able to manage the life support system, but will have no idea who won the last presidential election,” the developer adds.