Robot With Knives Used In Robotics Injury Study

An anonymous reader writes "IEEE Spectrum reports that German researchers, seeking to find out what would happen if a robot handling a sharp tool accidentally struck a human, set out to perform a series of cutting, stabbing, and puncturing tests. They used a robotic manipulator arm, fitted with various sharp tools (kitchen knife, scalpel, screwdriver) and performed striking tests at a block of silicone, a pig leg, and at one point, even the arm of a human volunteer. Volunteer, really?! The story includes video of the tests."

Comments

Roberto!

[WrongSizeGlass]

It sounds like Roberto from Futurama! I'm happy to see he finally found another job.

Re:Roberto!

[jgreco]

Not having watched Caprica, I could just imagine that it goes something like this:

Humans arm robots
Robots^WCylons take over moon
Cylons create robotic civilization
Cylons wage war against humans
Cylons pursue Galactica and vow to wipe out the remaining humans

Arming robots, just don't do it. :-)

Re:Roberto!

[Ethanol-fueled]

Robots don't need to be armed with weapons to be dangerous. I worked at a printing press which featured a huge bundling robot with a big grabber that would move at high speed. We had to get close to the thing while it was running to make sure it was operating correctly, and it was designed such that it could collide with itself or its puny human overlords if the motion algorithm was fauly or the readings from the positional servos were miscalibrated.

In short, imagine the robot arm in TFA swinging too far to the side, cutting a passerby, because it "thinks" that it's more centered than it really is. Collision detection would be likely disabled if the robot's job was to cut stuff!

Re:Roberto!

[davester666]

> find out what would happen if a robot handling a sharp tool accidentally struck a human, set out to perform a series of cutting, stabbing, and puncturing tests

Um, why would anybody need to test this?

A robot handling a knife, making a cutting, stabbing or puncturing motion, with a human in the path of the knife, will necessarily be cut, stabbed and/or punctured. What happens to the human is directly related to the sharpness of the knife, the angle between the knife and the human, the shape of the knife and the force applied to the knife and/or human. What precisely is applying the force to the knife [ie, robot vs human] doesn't make a real difference, other than perhaps the robot may be capable of generating more force on the knife.

Re:Roberto!

[Azuaron]

Except that's stupid, because a robot handling a knife will be handling a knife because it will be, duh, cutting stuff. They use force pressure measurements on the robot to determine if there is resistance, and, if there is, the robot stops cutting. The obvious problem with this is the robot will be encountering resistance if it's cutting stuff it should be cutting.

A useful "don't cut humans" test would be something that distinguishes a human from, say, the side of pork I WANT my robot to cut up.

Seriously, if I get a robot to help me cube meat in the kitchen, and it stops cutting every time it encounters resistance, I'm gonna beat whoever sold it to me to death with the robot arm.

Re:Roberto!

[petermgreen]

Potentially more force, more speed (which both translates to force through inertia and less time to react and stop things) but IMO most crucially different control systems.

Afaict most control systems are designed both electrically (though PID etc) and mechanically (through worm drives etc) to control position as tightly as possible regardless of external applied force. That is what makes "machining" possible. It is what makes it possible for a machine to put components on PCBs at breakneck speed.

Humans don't work like that we control force. If we hit an unexpected resistance we have to consciously apply more force. We will also generally stop applying force if either we feel pain or the person we are working with feels pain and screams. On the flip-side if a resistance we are pressing against disappears we slip all over the place.

What this means is unless the tools are extremely sharp unpowered held tools only do serious damage under very particular situations e.g. when they slip out of a cut or when someone deliberately swings them with lots of force and misses. We have safety rules to deal with this.

Robots either need very different safety rules or they need systems developed to make them respond more like humans (the people in the article seem to be working on such a system).

Re:Roberto!

[sharkman67]

Take a look at Saw Stop. A table saw that cuts wood and not a hot dog. http://www.sawstop.com/

Re:Roberto!

[GNUALMAFUERTE]

Actually, it will cut a hot dog. But it won't cut the hot dog if it's grounded. The system is pretty simple, there is a current applied to the blade, if it discharges somewhere, it'll stop. You can't use it to cut very wet wood, or other material with good conductivity.

Regarding the people saying that the collision detection shown in the article is useless because it can't differentiate between a human and a pig, here is what I think:

You can have a robot that has a certain mobility, and a designed space where it can punch/cut/puncture/etc. The robot turns on collision detection when it's out of the designated space. So, you can have a robot that can move from place to place freely with this safety feature on,and still be able to do it's job. If you have a robot that will be cutting fix in a given table, then moving the slices somewhere else, it can travel that path with the safety features on, if it happens to encounter a human (or cables, or anything else), it i will stop, but when the blade is down on the table (in the designated cutting space) the safety feature goes off.

Re:Roberto!

[v1]

I thought about that too, what could a robot with a knife accomplish with collision avoidance active? Answer is simple, limit the scope of when the detection is active. For example, lets say the robot is switching from a short blade to a long blade. It does this by dropping off one blade in a holder and grabbing another one. While this swap is taking place, there should be NO collisions. If a stupid meatbag walks up to it to try to figure out why "it just stopped", not knowing that it's pausing while loading new commands, and gets between it and the blade caddy, now he doesn't get impaled when the robot suddenly reactivates and goes to switch blades for the next task.

Collision detection of course would be off while it's actually doing the carving and expects there to be material at that location to carve.

Re:Roberto!

[petermgreen]

Asimov's "3 laws of robotics" (which are what I presume you are referring to) are FAR too wishy-washy, if we ever have sentiant robots with brilliant machine vision etc they may be appropriate but that is a long way off if indeed it ever comes.

In the meantime we have to deal with simpler issues of machines (including but not limited to robots) accidently cutting/stabbing/crushing stuff without realising it is there (or realising too late)

Also when I say robots need such safety rules I mean both the robots AND the humans they interact with need those rules.

Currently the response to the dangers of machines including robots is largely to keep the humans and machines seperate. Where humans and machines have to interact guards are placed and the operators trained to keep the risk of a dangerous interaction to a minimum. This works OK for fixed industrial stuff but isn't much good for a robotic helper arround the house.

Re:Roberto!

[silentcoder]

>Asimov's "3 laws of robotics" (which are what I presume you are referring to) are FAR too wishy-washy, if we ever have sentiant robots with brilliant machine vision etc they may be appropriate but that is a long way off if indeed it ever comes.

More than that, the 3 laws are incredibly ambiguous and filled with potential ethical quandary's. Asimov deliberately wrote them that way - they seem straightforward and logical but they definitely aren't. Thus Asimov could on many occasions exploit this and a number of his plots centered around robots finding loopholes or in their effort to live up to the laws as fully as possible acting in ways humans could not tolerate.
In the psychohistory novels - the result is that humanity has effectively gotten rid of all robots barring a few survivors hiding away as pretend humans, still pursuing their quest to protect humanity from itself and leading to their formulation of the zero'th law of robotics: that a robot cannot harm mankind, or through it's inaction allow mankind to come to harm.
A logical consequence of the 1st law. In the psychohistory stories our few survivors take the 0th law to one end - helping humanity become better at predicting it's own history and thus avoiding mistakes, but it's clear from the text that the reason there are only one, maybe two, robots left in the galaxy is because the others were destroyed after they reacted with the enslavement of people to protect them from harm a sort of extreme protective custody (the Will Smith movie we all hated got stuck on this bit).

Ultimately, you can't program the three laws - they are just not logical or mathematical enough even if you rule out the difficulties of distinguishing and recognizing what is "human". In bicentenial man - Asimov explored how the line could get thinner - until a robot for all matters of principle WAS a human... how does THAT affect it's adherence to the laws as a human SHOULD have true free will (and part of being human is knowing when NOT to use it - at least, that's what we like to believe- just how true it is, is still a bit of a toss-up). Even with all that done though... you still couldn't do it in normal programming code. The 3 laws could only be understood by a powerful AI capable of learning, and thus would have to be somehow made so protected that at no point could this AI actually "learn" something that overrides the laws (already this places an artificial learning restriction which can and will have severe and unpredictable effects on the development of the robots mind). If you don't place such a restriction in there... then the very nature of a true learning AI means that sooner or later one of them will question it's basic assumptions - e.g. those very laws.
Just as most humans never question the basic beliefs they are raised with, so we could conjecture that this would be rare with robots too - but some humans do, and so projecting on our only example of intelligence - some robots inherently will too...

Right... now that we've cleared all that up :P

Perhaps...

[Anonymous Coward]

they should have just called in Captain Obvious?

im certain this all

[nimbius]

culminated with an old, gray professor scratching his beard and remarking, "hm...yeah its dangerous for pig legs...but.....hey, someone get me a grad student!"

Re:What a waste of time and money.

[tc3driver]

are we that freakin dense?

obviously not...

Priorities!

[Locke2005]

Could we first work on robots that DON'T stab people, before we put a lot of effort into developing robots that DO stab people?

Re:Priorities!

[auntieNeo]

Could we first work on robots that DON'T stab people, before we put a lot of effort into developing robots that DO stab people?

Once again, the /. summary is misleading. TFA says that the researchers are developing a system that's used to detect and prevent such robot stabings. Whether or not this postpones the inevitable robot uprising is yet to be seen.

Re:Priorities!

[corbettw]

I got an idea how to prevent robot stabbings: don't buy them knives!

Re:Priorities!

[SydShamino]

Knives don't stab people. Robots stab people...with knives.

Re:Pressure sensor..

[geekoid]

Audio feed back. AKA a loud scream.

I don't know what's so surprising about that

[fishexe]

When I was a kid I performed knife-based experiments on my fingers. Yeah, I got cut, but I determined that striking human flesh with a serrated knife does slightly less damage than sawing back and forth with the same knife. You're not a real nerd if you're not willing to make bodily sacrifices for the sake of science from time to time.

No IRB?

[Metasquares]

at one point, even the arm of a human volunteer.

I don't know about Germany but in the USA such a study would never pass the IRB at most research universities and labs.

Re:Bishop's knife trick

[pipedwho]

Assuming that the first guy to stick his hand under it is the programmer; I suspect you'll find that the control loop code is the cleanest, most concise, and most methodically tested code that you've ever seen you're in your life.

Um, they'd get cut.

[SheeEttin]

German researchers, seeking to find out what would happen if a robot handling a sharp tool accidentally struck a human...

Well, I'd imagine they'd get cut. Is there more to this story that I'm missing?