Attack of the Drones
A few weeks ago, I posted an article about the kids using "FPV systems" -- virtual reality goggles slaved to drone cameras -- to remotely pilot drones through extreme aerobatic maneuvers. I said that they were the leading edge of the new paradigm for fighter pilots, training themselves for jobs that don't yet exist but will one day very soon. Well, it appears they may be out of a job before they're hired.
A PhD candidate at MIT has created an autonomous drone which uses neither radar nor any other form of active emissions for obstacle avoidance. Instead, it "sees" in visible light wavelengths and identifies obstacles by analyzing the video in real time, then operates the control surfaces in order to avoid the obstacles. Here's what the drone "sees":
Perhaps more amazingly, this is all done with off-the-shelf components. Two quad core microprocessors serve as the "brain" of the drone. If this can be built with what amounts to parts from Radio Shack, imagine what a big-budget DARPA program might accomplish. After all, it's only a small step from obstacle avoidance to visual target identification and engagement.
Why is that such a big deal? Because it means the drone has no active emissions. From an EMCON (Emissions Control) perspective, the drone remains virtually invisible as it seeks out, targets and destroys the enemy -- all at the speed of a computer. A squadron of these hypothetical attack drones could independently destroy entire wings or battalions of enemies before they even knew it was there -- and do it without human intervention. In fact, having a human actively in the loop would just slow it down. Watch the video at the link and you'll see how quickly it reacts to visual stimuli and executes control inputs.
When I was a boy, I read the series of science fiction stories written by the late Keith Laumer which centered around the Bolos: self-aware and highly intelligent "supertanks," any one of which was supposedly capable of wiping out an entire conventional army. The largest, which Laumer called "continental siege units," were dropped from orbit to secure entire enemy-occupied planets. The premise is that the heavily-armed Bolos react so quickly and think so far ahead of the enemy, no mere human force could even get near them, let alone destroy them.
If you'd asked me a year ago, I'd have said that Bolos were centuries or even millennia away from becoming reality. So much for that quaint idea.
The one thing you can be certain of is this: it obviously can be done -- right now, today -- and so it will be done. It's simply a matter of who does it first. Being an American, I'd prefer it was us.
A PhD candidate at MIT has created an autonomous drone which uses neither radar nor any other form of active emissions for obstacle avoidance. Instead, it "sees" in visible light wavelengths and identifies obstacles by analyzing the video in real time, then operates the control surfaces in order to avoid the obstacles. Here's what the drone "sees":
Perhaps more amazingly, this is all done with off-the-shelf components. Two quad core microprocessors serve as the "brain" of the drone. If this can be built with what amounts to parts from Radio Shack, imagine what a big-budget DARPA program might accomplish. After all, it's only a small step from obstacle avoidance to visual target identification and engagement.
Why is that such a big deal? Because it means the drone has no active emissions. From an EMCON (Emissions Control) perspective, the drone remains virtually invisible as it seeks out, targets and destroys the enemy -- all at the speed of a computer. A squadron of these hypothetical attack drones could independently destroy entire wings or battalions of enemies before they even knew it was there -- and do it without human intervention. In fact, having a human actively in the loop would just slow it down. Watch the video at the link and you'll see how quickly it reacts to visual stimuli and executes control inputs.
When I was a boy, I read the series of science fiction stories written by the late Keith Laumer which centered around the Bolos: self-aware and highly intelligent "supertanks," any one of which was supposedly capable of wiping out an entire conventional army. The largest, which Laumer called "continental siege units," were dropped from orbit to secure entire enemy-occupied planets. The premise is that the heavily-armed Bolos react so quickly and think so far ahead of the enemy, no mere human force could even get near them, let alone destroy them.
If you'd asked me a year ago, I'd have said that Bolos were centuries or even millennia away from becoming reality. So much for that quaint idea.
The one thing you can be certain of is this: it obviously can be done -- right now, today -- and so it will be done. It's simply a matter of who does it first. Being an American, I'd prefer it was us.
posted by Jar(egg)head at 13:47
1 Comments:
Driverless cars, remote self piloted aircraft, just name the first one T-1 and we shall be the architects of our own destruction. Please just keep Skynet away from these things.
Post a Comment
<< Home