The success in the robot race of DARPA is another step on the way to autonomous combat robots
Robot have been in the modern war for years. The cruise missiles of the american and russian dispute are nothing more than selfishly targeted bombs, which penetrate deep into the area of an opponent and accurately detonate their deadly freight at the desired place by means of modern GPS and radar-assisted landscape tracking. So to speak kamikaze robots. This type of robot is ultimately stupid, the target coordinates are previously determined and the robot flies on a predetermined route – smaller deviations corrective – in the desired target.
Another example is the unmanned declaration aircraft, so-called UAV (unmanned aereal vehicle), which are practically remote drones, but also take certain tasks. You can achieve a previously entered target area automatically and then stay in the air in a high up to 30 km until you are retrieved for certain tasks. Escaped with steering crakons you can even attack goals. But only with an officer in the background, in safe distance, sometimes even from a usage center in the USA, the actions of the UAV on the screen and granted instructions to a control panel (trained pilots or computer players?To).
Stanley, the VW-touareg des stanford racing team, is the winner of the DARPA competition grand challenge. Image: stanford racing team
Thus, we have to do it with a further, far more important step to the automatic war: the robots meet in-house decisions. Although this is reflected in relatively simple tasks, such as self-proposed driving in unknown countries. Soon, however, the border was allowed to be over-wound up-to-standing and sliding combat robots.
First, however, it’s about the problems of the replenishment. Because of the high losses of human life on the subsequent skills, the american congress has commissioned the pentagon to leave about a third of all land vehicles from the militar of robots by 2015. Last week now, the DARPA (defense advanced research projects agency) in the mojave-weste has organized the second robot race of the world under the title grand challenge.
Race in the weste
After none of the robots over a few kilometers have come out last year, there was the first successes this year. The vehicles had to have an obstacle distance of about 212 km within 10 hours of self-standing. 4 robots totaled 23 have done it. The winner of the race was stanley, a converted volkswagen touareg of universitat stanford.
As for the winner, the winner is different from most others that it is a stain-suitable family vehicle that was only slightly modified. Stanley has some greater tire as a usual VW touareg, a stompied stobstick and a metal plate for the protection of the underbody. In addition, control, brakes and accelerator pedal were coupled to the computer control. The strategy for autonomous driving is actually very simple, say the makers on their website:
The robot must know where he is to recognize the way, the best route elections and act afterwards.
A variety of sensors together with a GPS system determines the position of the vehicle to about 5 cm exactly. Laser sampling devices on the roof of the vehicle continuously follow the nearer environment of the vehicle. They pay particular attention to obstacles, such as stones, pothole, barbed wire fence or other objects, for example on vacant competitors.
However, the laser scanner can only look for about 4 vehicles. At speeds of 50 to 60 km / h, as required by the race, the vehicle therefore uses a radar system, which continuously scans the other horizon continuously. Cameras mounted behind the windshields – because of the risk of dirt sprayers – draw up high-resolution stereo images of the way. All incoming information is then processed simultaneously in several computers. The computer algorithms "learn" know the terrain and determine the optimal route.
"H1ghlander" the carnegie mellon university (CMU) was second. Image: cmu
"This is actually a software competition," says professor sebastian thrun, head of stanford team over the race in the weste. Several monitors in the wearer show the complicated "thinking process" of the robot, if he analyzes and weighted all tenth-seconds the relative safety of the various possible routes. A probability algorithm of the senior software developer michael montemerlo interprets the FULLE of the incoming sensor data.
Overall, stanley has 6 pentium M laptops, which are housed in the trunk in a shock-proof housing, to survive the adverse conditions on the custodian route. The computers are merged into a network. All programs are available several times, so even in case of failure of one or two computers, the vehicle still remains manovrier and decision-making.
The economical turbo diesel of the touareg allows the vehicle to survive the entire 10 hours of race with a single tank finishing and to provide even the entire on-board electronics.
Since the touareg is fully durable and officially approved as a dust vehicle, the team of the team could produce prof. Thrune to sit in the car every free weekend and drive to one of the off-road bands outside of los angeles, to then "let stanley" let go ". After the test drives, the employees often met in the night hours in a bar for off-roader to further optimize the programs.
"We have practically supposed a corner of the bar, unfolded and programming our laptops," said gabe hoffman, a doctoral candidate of the aircraft and space institute, who worked on the control software.
Many of the scientists and students involved in the robot race race and student traumas from the use of their vehicles with artificial assistants. Especially the members of the CANJUNBOT team, the university of louisiana. They had spent the last days before the robot race to save people from the destructed new orlean with their dinghy.
Of course, this robot technology will eventually be used to help people. However, these machines are first developed once to make people better and above all more efficient to death.
The winners of DARPA competition. After stanley and h1ghlander sandstorm, also from the CMU, the third place. Image: gizmag
Impact on the war of the future
The race in the weste has shown that it is feasible to drive robot in a self-sufficient in unknown terrain. First, we can ame that in the coming years, the autonomous moving vehicles can be significantly improved. In about 5 to 7 years, it will give first for the militar application-made systems. These are then used – as currently planned – especially for the supplies.
Since the follow-ups were still being attacked, it should be foreseeable that the robots soon also receive small weapons for self-defense to defend themselves against the attacks. Thus, it is no longer a way more to the fully automatic tank, which protects the supply vehicles, but can also be used for attacks.
For a long time, the pentagon has been working on visions for automatic war carrying. Especially in times of tight recruits, such visions have their charm. In addition, in today’s highly specialized army, people can only be difficult to replace. Robot, on the other hand, can be multipled arbitrarily, especially when they are produced in fully automatic production stations. And finally, the inhibition threshold becomes considerably reduced to the dead if the militars of the future only have to print on a button and then the robots then do the dirt work alone.
In addition to fully automatic tanks, which should turn off in a first attack shaft especially enemy tanks and artillery, in the arsenal of the arms forge is already working on other perfidic combat robots. So there has long been trying to make mines smarter and more mobile (intelligent mining network). Instead of simply waiting for the mines to drive a vehicle over them, such robot mines should actively look for a target and then attack it right away. There are already first prototypes that crawl like small insects and have a scoring radius of several hundred meters.
X-47: the next generation of an armed drone. Picture: DARPA
The air force has been working for years of developing fully automatic fighter aircraft. Obviously, the UAV (unmanned aereal vehicle) in the iraq war for declaration tasks have already preserved unexpectedly, (the iraqi airspace is full of drones) so you want to automate artfully bombers and fight luggage. This has two advantages: on the one hand, the training of a combat pilot is extremely expensive, tedious and expensive. On the other hand, man in today’s highly complex combat aircraft is more and more the vulnerability. Already a pilot can not fly a modern fighter airline without supporting computers, let alone fight. In addition, the acceleration faith, which occur in case of fighters, are carried out at the border of the for humans.
Already in the nineties, the british air force in experiments was taken to overcome fight pilots in their simulators against neuronal networks. Neuronal networks are, so to speak, the next step on the way to artificial intelligence. They do not work like a computer, but simulate the function of biological brains. If these neuronal networks were not even able to fly self-standing at first, so they have learned quickly and could be practically no longer defeated after thousands of virtual combat flows from the british pilots. Conversely, they were better and faster than any combat pilot, because they have much more short reaction times, which can process tons of data at the same time, do not have physical boundaries in just the acceleration faith and maybe because they are not afraid.
In addition, there are further achievements of the warmaster. In particular, the prospects on nano-robot. These are smallest machines – smaller than a housefly – which is in tangible near the significant progress of nanotechnology in tangible close. Here, the fantasy of war strategists already knows almost no limits: they dream of millions of smallest attack rods, which specifically distribute biological or chemical warfare, penetrate into the body of (human and mechanical) opponents and place them lame or just spy on the opponent’s general staff of the opponent. So far, such ideas still fail at the lack of miniature energy supply for the nano armies, but that too can change soon.
Another popular field of murder strategists is space, you need the overview of the overview of the robot armies of the future. Without functioning satellite communication, it is difficult to keep control of the combat machines. It must be prevented that the combat machines are off each other or worse, attacking their own people. The already mentioned GPS system as well as dozens of other communication and monitoring satellites must therefore be used under all circumstances. The robot war is therefore also the space war like a sibling.
Even for the home protection, the modern war robots can be ideal. In the short break, the american homeland security department has shown interest in procurement of a hundred UAVS. These are then to be the border with mexico and especially the extensive US cuffs fully automatically monitor.
There it is then no longer far to monitor your own population. Thus, starting locks with robots in the air and on the strain would like to control practically lubrication. Even better, such technologies are suitable for guarding prison camps or boundaries, the technologies for this are practical already use. This also has the advantage that the shoot command is fully automatic. Nobody needed to have more conscience conflicts. And another plus: human failure or "rubbing" were excluded. Too bad that honecker may not live anymore. The goods jumped into the air before joy.
The gladiator, an unmanned fight robot of the U.S. Marines. Image: pentagon
War of dumbbonds
Ultimately, however, these "thinking" robots are still plenty stupid. You at best reach the intelligence of insects. All these robots "decide" after a predetermined program. If situations occur, which were previously not considered or the for the system is new, the robot can often no longer react properly. The comparison with the insects is definitely to take. Just like a moth flying around the light, because it "thinks", she flies towards the sun, robot often behave as soon as something does not work as intended.
A good example could be seen in the robot race of last year. With one of the robot vehicles, the GPS system was defective. In the episode he turned to his own axis and did not come from the spot – the moth let pits. Here is also the chance for the art opponents of the robot armies. Often already with simple tricks, the sensors of the machines can be drilled in the wrong.
A popular game was already in the iraq war the manipulation of the GPS signal. With relatively simple means, the GPS signal grid can be moved with a storing transmitter by a few hundred meters. This is often enough to confuse simple robots and let them get off the way. Of course, however, there are again counterparts. The newer cruise missiles therefore use very detailed landscape matches for the destination approach, but also you could do it with home cottages a la hollywood.
Battle robots of the future will be very strong on infrared sensors, as every person and also all machines radiate warm, so that important of unimportant objects can be well differed. And especially at night, together with remaining light stronger, they have a significant advantage against the opponent, which is practically blind without this technology at night. In addition, the robot can then locate the mouthfire of the attacker very accurately because each shot as a small flash of light in the infrared area is clearly visible.
However, such infrared sensors are simply dazzling through a variety of small fires. Or you could confuse the combat machines through small flashing infrared leds, simulating the merd-fired attacker. Simple bike indicators could even enjoy this purpose. If, on the other hand, aim for the infrared detectors of the combat machines, one must probably do not go into a bullet-proof vest, but in a styrofoam suit agony. Also consideration value ware against the radar sensors with built-in boiler, which minimizes the hot picture of artificial soldiers. Or the opponents simply capture one of the friend enemy recognition systems unnoticed, with which the robots are all equipped. Then they could safely met the fight robots and turn them into turn.
Another weak point of the machine is your connection to the command center. You will not send such robot to the combat area without a last human influence. That turned the opponents to take advantage of. Art generations of hackers could turn off whole armies or even judge their own people.
A wide field of war fantasy are also future anti-robot robots. It was also allowed to work with high prere in the high-tech weapons forests in the world. Because what could better get a robot than another robot? So here is ultimately a new bettrust in motion – with a prospective view.
Factory leaflets of whole population groups
Above all, the development is not limited to state-funded murder commands, everyone can soon order their own combat robot on the internet (remote-controlled terrorist attacks). Already there is a rough community of fight robot enthusiasts. These are trail of remote-controlled combat machines that compete against each other in television shows. For a few hundred euros you can get the latest combat machines on corresponding websites. Soon there will be the appropriate software for self-sufficient operating systems.
Already there are first neuronal networks in striking electronic shops against bares. Even computers that can keep up with the fastest of the world consist only of connected play station motherboards. Every hobby hobbyer can soon build his own cruise of missiles, reach perhaps 5 to 50 km, that is enough for most applications (cruise missiles in self-construction). I’m already wondering for a long time why it actually gives suicide attendants. Do not have access to conrad electronics?
Therefore, fight robots are far from what the commutations want to make us: an option to avoid human victims in the event of a war. Rather, that meets the opposite. There is a risk that the war is expanded by a terrible dimension: the factory-maby leaflets of entire population groups.
There will be a development, with a completely uncertain output. Only dictatorships can be pleased with these prospects without exception, they no longer need the human factor to fear. There is also a rough danger for artificial wars. Until now, the soldiers involved had to be won by skillful media propaganda for the use of their lives. We are experiencing that again using the example of iran. In the fully automatic war, this barrier is also falling.
And another point: in automatic tuning machines, civilian victims will not be avoided. Although combat machines can recognize their own dispute with relative safety through modern friend enemy recognition systems, all others are then potential enemies. In a wiser anticipation, however, the pentagon has already provided. A first important step for the automatic war has already been done: the number of toten civilians in combat actions is no longer paid by the pentagon for several years.
There is also a already word for these victims of military progress: collateral damage, unintentional accompanying damage then does this on neusprach.