Stanford Students Build "JediBot"

An anonymous reader writes "By combining a dexterous robotic arm, a foam-padded lightsaber, the movement tracking capabilities of Microsoft's Kinect sensor, and some clever software, students at Stanford University have created what can only be called a JediBot. Using a series of pre-programmed 'attack moves', and Kinect to detect the location of the enemy's light saber, JediBot can attack and defend with surprising grace. For now its attack moves are fairly slow — it can only attack once every 2 or 3 seconds — but presumably you could tweak a knob (and remove the foam padding) to turn JediBot into a real killing machine." I look forward to model that can also "force choke" an opponent.

Comments

its all fun and games

[Osgeld]

till the bot spazzes out and whops some one ... then its just fun

Grievous Mistake

[Greg Merchan]

This seems to me to be a grievous mistake.

This is only a toy

[Okian Warrior]

As a long time swordfight student and instructor, I have to say that putting a stick in the grip of an industrial robot does not make it a sword fighter.

For example, the obvious mistake seen in the video is that the robot strikes at the *weapon*, not the *person*. As any 1st year fencing student knows, you can't win the fight that way. I love fighting nubies who make that mistake.

We have a term for this - it's called "Erroll Flynn" fighting, and it refers to those cheap movie swordfight scenes where the actors aren't skilled enough to actually fight without putting out their opponent's eye. Stay far enough away so that you can't hit the opponent, and cross swords in mid air. Clack... Clack... Clack... now low: Clack... Clack... Clack... now high...

Let's have a robot that holds a broom and say it's a sweeping robot! Or a robot that holds a hose and say it's a car washing robot! Or a robot that holds a trimmer and say it's a hedge-trimming robot.

Wake me when it can detect an opening in the opponent's defense and strike at it.

Re:This is only a toy

[gclef]

In its defence, this is exactly what the movie jedi did also, so calling it a "jedi bot" is an accurate description.

Re:This is only a toy

[diewlasing]

I'm not sure if you're joking or you really did miss the point of the experiment. Yes, obviously a more realistic robot would try to actually attack the sword. But this isn't the main point. The point is to see if you can actually engineer a robot to respond to different situation appropriately. And they did. I have a have a ladder that goes great with that high horse of yours.

Re:This is only a toy

[nedlohs]

Wake me when it can detect an opening in the opponent's defense and strike at it.

I suggest you don't bother reading slashdot. If a technology has to be at that level beefore you want to even hear about it then maybe you should get your tech news from the History Chanel?

Re:meh

[NFN_NLN]

Why not build a better one yourself then?

Someone beat me to it.. in 2009:

http://www.hizook.com/blog/2009/08/03/high-speed-robot-hand-demonstrates-dexterity-and-skillful-manipulation

Re:Force Lightning

[TheRaven64]

Force lightning can be directed. You'd need something that would ionise a path through the air to act as a conduit. This has been done before with lasers. It was originally intended as a stun weapon - like a taser but without the need for wires - but the laser burns tended to do more damage than the electrical arc.

Re:Laws of Robotics

[TheRaven64]

Why not? Pretty much every robot produced can violate the first law, because no one has created a control system capable of defining what 'harm' means. Even toy robots can move just outside of a baby's reach and make them cry. Most factory robots could probably disassemble a human if one got in the way.

Most of Asimov's robot stories dealt with the fact that 'harm' is difficult to quantify even for highly intelligent beings. For something with a simple control program, it's basically impossible.