Here it is, the ultimate toy-and if you break it you can't get caught
HERE'S a moral dilemma to mull over this Christmas. You're playing with a model train set, not yours. You're bored. What you'd really like to do is put one train on a collision course with the other, sit back and watch the disaster unfold.
But the collision might damage the trains. More worryingly, the owners might catch you at it. And how would you feel if this were your train set?
Relax. You're controlling this railway over the Internet. A live video feed shows you the trains and a program allows you to pick the destination of each one. You don't know the owners and they don't know you but, by making the website public, they have invited you to play. If they've placed two trains on the same track, that's their problem-and anyway, you could claim that any collision was simply an accident.
Welcome to the world of telerobotics, the emerging field of long-distance remote control. Forget the comparatively primitive kinds of remote control we use to switch TV channels, say, or steer radio-controlled cars. For teleroboticists, the vision is far grander.
Their world has been transformed by the convergence of two technologies: the Internet and robotic machines. It means that one day, schoolchildren could carry out experiments on the space station while engineers on one side of the planet fix a broken power generator on the other side.
But this revolution could also change our notions of responsibility and morality. Punch someone and they'll punch you back; crash your car into someone else's and you pay to have them fixed. This knowledge keeps the world in a delicate balance. Change it and who knows what will happen? The dangerous thing about telerobotics is that cause can be separated from effect.
But don't head for the bunkers just yet. As with any emerging technology there is fun to be had while it matures, and the nature of telerobotics means we can all play. Want to drive remote controlled cars in the US or control "remotebots" in Germany? They're all there on the Web (see "Click here to play"). One robot, called ANU, at the Australian National University in Canberra can be steered through the campus offices. Naturally, the offices are full of people, conjuring up a scenario rich in potential accidents-a bruised leg or a crushed toe, perhaps. Go on, they'll never know it was you.
Then there is the bot that moves among the snakes in a Brazilian reptile house. The images send a chill down the spine but so does the possibility that your instructions to the robot could be crushing a rare and beautiful animal to death. It might have been a genuine accident this time but try telling that to the zookeepers. Who would believe you after what you did to those trains?
Of course, the destructive potential of these robots is not quite that easy to harness, at least not yet. For the most part, telerobots are clumsy and unreliable, and as for the Web, anyone with a standard domestic Internet connection knows how well that works. "The major challenge is overcoming the time delay the Internet introduces," says Roland Siegwart, a teleroboticist at the Swiss Federal Institute of Technology in Lausanne.
Log on to steer a robot and you'll see what he means. Your commands take time to reach the robot and the images take even longer to come back. The return journey takes at best a second or two and at worst forever. It doesn't take much imagination to foresee the problems this causes.
NASA knows this better than most. In 1996, it sent a rover called Sojourner to explore the surface of Mars. The robot was given some autonomy, but to prevent accidents, NASA built in safeguards that made the rover stop and wait for commands from Earth if, for example, it started to tip over.
What NASA had not anticipated was that the rover would constantly tilt beyond the fail-safe point and be forced to wait ages for a message from Earth to set it going again. As a result, it was stuck in an area called the Rock Garden for weeks.
That's not a problem for the latest project from Ken Goldberg, an electrical engineer at the University of California, Berkeley. He has experimented online with everything from gardening robots to Ouija boards. Now, for the ultimate telerobotic experience, he has connected a person carrying a webcam to the Internet and allows users to control the person's movements.
Goldberg says this could revolutionise the traditional school trip. A "tele-actor" could travel to the rainforests or the top of a mountain while the school party decides from the classroom which way to go. Turn a journalist into a tele-actor, and television news might never be the same again.
Goldberg worries about the moral issues that telerobots raise, so he set out to test them. In 1996, he set up a telerobot that allowed users to burn holes in a $100 note, which is a crime in the US. How many users would think twice about such an anonymous and apparently harmless crime?
But the exercise was more complicated than that. One claim was that the website was faked. Instead of live images, the suggestion was that users were seeing a series of carefully selected library pictures.
This issue puts telerobot users in a bit of a pickle. How can they ever know that their commands are really carried out? And if so, why should they worry about the consequences? According to Goldberg, this is an insidious problem that could eat away at the moral fabric of an online society.
Imagine, they say, a website showing live film of a human head in the cross hairs of a rifle, with a button that controls the trigger. Pressing the button shows the head blown apart by a bullet. If you pressed the button would you be guilty of murder, attempted murder or just plain stupidity? Would you even care?
Thankfully, not everyone wants to spend their surfing time taking potshots at strangers. So if you aren't lucky enough to get a train set this Christmas, no matter. Just log on and start playing with the online model. And while you're there, be thankful that each train is programmed to move out of harm's way should some idiot try to smash them together.
From New Scientist magazine, vol 172 issue 2322, 22/12/2001, page 62
© Copyright New Scientist, RBI Limited 2001