Unmanned, remotely operated military drones are either loved and hated in almost equal measure, depending on who you ask, of course.
But the U.S. Air Force is ready to move on to the next step: unmanned, self-aware drones.
So if you are one of those people who always dreaded the possibility of robots taking over and decimating every living thing on the planet, well, that day may have just drawn a little closer.
Yes, the new drones are not only expected to think, but to also anticipate the movements of aircraft operated by humans.
But before anyone goes into a panic, this project was not conceived out of a strong desire to actualize robot-led Armageddon. Rather, it addresses a legitimate concern – that of drone landing and takeoff at busy airports.
There is a real risk of a drone colliding with manned aircraft simply because changing the course of a drone takes significantly longer than the reaction of a human pilot.
When confronted with danger, a human pilot acts instinctively and swiftly, something that can make all the difference between a collision and a near miss. The goal is thus to create an algorithm for drones that would effectively "read" the minds of a human pilot flying too close to a drone.
A number of companies have already proposed some interesting ideas. Soar Technologies, for example, intends to code an algorithm that relies on pattern matching, memory management, goal based reasoning and cognitive explanation mechanism. The last feature will enable the drone to detect when a pilot’s flight pattern is erratic or out of control.
On the other hand, Stottier Henke Associates, a California based company, asserts that a single algorithm is unlikely to capture all the functions necessary to predict pilot intent. However, Stottier has yet to reveal exactly how it intends to implement a multi-algorithmic approach to realize this goal.
[Via Wired Danger Room]