Thinking outside the box is that they do – really outside the box
Isabella Mayer de Moura, with information by Darpa
The United States defense research agency which took the first step towards the creation of the internet has an impressive history of innovative, curious, and, well, pretty bizarre projects.
It is the Defense Advanced Research Projects Agency, known as DARPA, which appeared over 60 years ago as an American answer to the launch of Sputnik, the first artificial Earth satellite, by Russia. Ever since, projects by their researchers originated innovative technologies, such as precision weapons, small GPS receptors that were incorporated into cell phones, voice recognition, and automatized language translation systems.
Money doesn’t seem to be a problem for this agency of the US government. With 220 employees, DARPA has a budget of more than US$ 3 billion, and is expected to be granted US$ 3.5 billion in the 2020 fiscal year – an amount which is yet to be approved by Congress.
With that money, the agency has over 250 research and development programs. Here are the most important technologies which may be generated by these projects.
Controlling a weapon with one’s mind
Artificial intelligence for military purposes has been one of DARPA’s focuses. In the next five years, the Pentagon agency will invest US$ 2 billion to develop what it has been calling the “third wave” of artificial intelligence projects, catering to the growing appetite of the military for technology.
One of the projects to be developed from this year on, in a partnership with the private sector, is the ability to use artificial intelligence in order to control machines telepathically.
Through a program, authorities want to build artificial intelligence into neural interfaces – a technology that would enable a person to control, feel and interact with remote machines as if they were a part of his own body.
These neural interfaces have been used to allow people to control prosthetic members, translate text thoughts, and control drones remotely. Through the Intelligent Neural Interfaces program, DARPA will explore how artificial intelligence can make these systems more durable, efficient, and effective.
The teams taking part in this project may receive financing of up to US$ 1 million, and will have 18 months to build the prototypes.
Adaptable artificial intelligence
The current artificial intelligence systems are excellent in tasks defined by strict rules. They are able, for instance, of dominating the rules of a chess game, and overcoming the best human players. They are not very good, however, in adapting themselves to changes which are usually faced by army troops in the real world – whether it is a real surprise action by an opponent, weather conditions, or operations into unknown territory.
“For artificial intelligence systems to effectively associate themselves to humans in a specter of military applications, intelligent machines need to upgrade from solving closed-world problems, within confined limits, to open-world challenged, characterized by fluid and new situations,” says DARPA.
With the same goal, and along the same lines as the previous project, the agency released the Science of Artificial Intelligence and Learning for Open-World Novelty (SAIL-ON). In it, they plan on teaching to an artificial intelligence system how to learn and react appropriately, without the need to be trained in a big database.
“It would not be practical to try to generate a database of millions of kilometers for land military systems that need to travel off-road, in hostile environments, and are faced constantly with new high-risk conditions, and especially less for autonomous military systems which operate on air and at sea,” said Ted Senator, DARPA’s Defense Sciences Office program manager.
According to the agency, the program tries to establish the technical bases which would enable machines, regardless of the situation, to undergo the military cycle known as OODA – observe, orient, decide the best course of action, and then act.
This other project promises to aid in the treatment of war injuries, such as those caused by explosions, burns, and other wounds which usually cause catastrophic damages to bones, skin, and nerves.
Using biosensors and artificial intelligence, DARPA believes it will be possible to drastically improve results in the recovery of tissue injuries. Through the Bioelectronics for Tissue Regeneration (BETR), they want to follow up closely the progress of the injury, and then stimulate healing processes in real-time, to optimize tissue repair and regeneration.
According to Paul Sheehan, BETR manager, the technology would allow “not only a personalized medicine, but also dynamic, adaptable, and precise human therapies,” which would adjust themselves to the wound at every moment, to provide more resilience to injured combatants.
“To understand the importance of adaptive treatments that respond to the wound state, consider the case of antibiotic ointments,” Sheehan explained. “People use antibiotics to treat simple cuts, and they help if the wound is infected. However, completely wiping out the natural microbiota can impair healing. Thus, without feedback, antibiotics can become counterproductive.”
This new technology could, for instance, modulate the body’s immunological response to the wound, recruiting the necessary types of cell or indicating how stem cells should differentiate themselves in order to accelerate the healing process.