By KEAY DAVIDSON San Francisco Chronicle April 10, 2006
These machines won't be lovable-looking, clankety-clank robots like Robbie in the 1956 film classic "Forbidden Planet" or his similar-looking, near-hysterical cousin ("Danger, Will Robinson!") in the 1960s TV show "Lost in Space." Nor are the machines likely to look or sound like the red-eyed, suave-voiced Hal of the 1968 blockbuster "2001: A Space Odyssey."
Rather, these computerized "intelligent systems" could be near-invisible legions of microelectronic butlers and maids. Hidden inside the spaceship's computer circuits, they'll sleeplessly monitor the ship's sensors to ensure it isn't about to blow up, lose air pressure or veer off course. There's plenty at stake. In space, catastrophes can happen incredibly quickly, far too quickly for humans to take protective action (unlike "Star Trek," on which Capt. Kirk repeatedly evaded mishaps by barking orders at his crew). Intelligent systems will "think" at the speed of light, hence, they should have a better chance of reacting in time to save the day - in theory, at least. NASA is now developing the Crew Exploration Vehicle, a post-space-shuttle ship. According to President Bush's blueprint for the future of NASA, the new vehicle should be able to ferry humans back to the moon by 2020 and - maybe a decade or so later - to Mars. At NASA's Ames Research Center in Mountain View, Calif., researchers are busy trying to develop intelligent systems for the spaceship and its launch booster. The risks of Mars missions are immensely worse than anything faced by the Apollo astronauts of the 1960s. The moon voyagers were never more than 240,000 miles - a few days' flight - from Earth. Back then, if anything went wrong, they could scurry back home. That's what the Apollo 13 astronauts did after an onboard explosion, and all three survived. By contrast, a typical flight to Mars will last a few years. During that journey, the astronauts will be exceptionally vulnerable to micrometeoroid impacts, which could be powerful enough to burst open their spaceship; cosmic rays, which are high-speed solar and galactic particles that could sicken and perhaps kill them; mechanical failures caused by lengthy exposure to the severe cold of the cosmic void; and impaired judgment or psychological deterioration caused by their long isolation in a strange, hostile realm. "Life-support systems for Mars will have to be absolutely fail-safe and bulletproof. If the life-support system fails and you're four months (travel time) away from Earth, you're dead," said Steven Zornetzer, deputy center director at Ames. As now envisioned, intelligent systems will be exceedingly complex - which is itself a problem. The more complicated a computer - or any machine, for that matter - the greater the number of ways for it to malfunction. "The actual 'failure space' - all the possible things that could go wrong - is enormous," said Sandra Hayden, a software designer at Ames. Even if such a system is successfully developed, it will take a long time to test or "certify" it to ensure that it works reliably. For example, "if you double the complexity, you quadruple the (system testing) that needs to go on to make sure everything will go right," said Ames mechanical engineer Dougal Maclise. The same cosmic rays that could put astronauts in danger also could damage or destroy microchips in their spaceship's computers. One possible solution is "self-healing" microchips. Experts speculate about putting multiple copies of the same chip on a tiny superchip. That way, if one chip fails, the computer can route instructions to a properly functioning chip, says team leader Serdar Uckun of Ames. Another potential problem is that as computer systems grow more complex, so does their tendency to trigger false alarms. The problem of false alarms "is one of the biggest challenges we face. You don't want to abort a mission when you don't need to abort it," said Mark Schwabacher, an Ames computer scientist. To better the computers' odds against false alarms, NASA experts hope to educate them, in effect, by feeding into them data about malfunctions on other spaceship missions as far back as the 1960s. Unfortunately, past experience can be deceptive. For example, during numerous space-shuttle launches over the past two decades, insulating foam repeatedly fell off the shuttle external fuel tank. The incidents caused little or no obvious damage. Hence, few NASA officials regarded falling foam as a serious risk. But in 2003, falling foam damaged a wing on the shuttle Columbia. As a result, the spaceship and its seven astronauts burned up on re-entry. That leaves NASA computer programmers facing something akin to what philosophers call the "white swan" problem: If all the swans you have ever seen or heard about are white, how do you anticipate the possibility that one day you'll see a black swan? Because humans haven't figured out the answer to that one, it's anyone's guess how to program computers so that they can answer a similar question en route from Earth to Mars: Is that funny sound in the turbo-pump just a minor glitch, like all the previous funny sounds that proved to be harmless - or an omen of an imminent explosion?
Distributed to subscribers for publication by Scripps Howard News Service, http://www.shns.com Publish A Letter on SitNews Read Letters/Opinions
|