| SPACE ODYSSEY: The Mars Exploration Rover Spirit and its lander; (Below) ASIMO
In 1989, using an insect-like robot named Genghis, Rodney Brooks pitched a bold vision for exploring space: Send up an army of small, cheap machines to rove around on a distant planet and beam back data.
The concept kicked off a new era in robotics, and eight years later, NASA sent the simple probe Sojourner rolling across the surface of Mars.
But now Genghis sits in a box, and the sophisticated machines that populate Brookss lab at MIT are becoming increasingly human-like: One has a sense of touch, another can find a familiar face in a crowd. Eventually a robotic torso named Domo ' now learning to wield a screwdriver ' will be able to master new skills by imitating people instead of undergoing software updates.
The new designs are part of a broader shift toward a vision of robots that are partners, not simply remote-controlled probes.
The change has been fueled by more powerful computers and better robotics as well as by new space policy. The Bush administrations push for more human space flight ' signed off on a few weeks ago by Congress ' is increasing the demand for robot partners that can learn new tasks, use tools the same way people do, and act as a space support staff.
The thing we were tasked by NASA is: how can robots support manned missions on the moon and Mars before people get there, while they are up there, and after they've left, said Brooks, director of the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology. The danger is sticking with the mind-set that developed in the 1960s of what robots do and what humans do.
Cheaper and safer
It is now clear that both humans and robots have their advantages in space ' and the segregation between the two is fading as NASA pursues colonisation of the moon and Mars. Robots pose less risk: No one loses a parent, child, or spouse when a spacecraft carrying a robot explodes. They are also cheaper and less delicate explorers than their fleshy creators.
The Mars rovers, Spirit and Opportunity, for example, have cost roughly $900 million over 5 years, according to Joy Crisp, a project scientist at the Jet Propulsion Laboratory in Pasadena, California. In contrast, in its first 112 shuttle flights, the three-decade-old space shuttle programme has cost 14 lives and roughly $1.3 billion per flight ' not a great bang for your buck, said Roger Pielke Jr., director of the Center for Science and Technology Policy Research at the University of Colorado at Boulder.
But even the best-designed robots cant compete with the human brain
The rovers are an amazing success: Both are still working after two years instead of the projected 90 days. Theyve each put about four miles on the odometer, sent back possible evidence of water, and collected data to keep scientists busy for years.
They were never really alone, though. A ground team of about 50 scientists and engineers told them what to do each day. People mined everything the rovers saw, felt, and measured, and then had to tell them to go back to probe the interesting spots. A person could walk the same four miles in a single day, said Jeffrey Hoffman, a former astronaut who worked to repair the Hubble telescope in 1993.
So the real trick will be to blend the versatility of the human brain with the efficiency of the robotic mechanisms.
Let them do what they do best, look at it as a team, said Dava Newman, a professor of aerospace engineering at MIT, who studies the biology of humans in space. Youre right there with your robot assistant ' the robot is kind of the laboratory, helping analyse the soil, maybe helping carry the equipment.
Newman, like other scientists who believe that people should explore and live in space, realises that they will depend on machines as precursors and as complicated tools.
In Brooks lab, robots are developing the skills theyll need to be useful to people ' though he still believes that fast, cheap, and out-of-control robots will have a role in the future. The personable, twitchy robot Mertz recognises faces and distinguishes one person from another. Obrero, a mechanical arm, has a touch so sensitive it can pick up a fluffy stuffed cow or a plastic toy ' delicate tasks that have foiled many other robots.
At the Johnson Space Center in Houston, researchers are developing Robonaut, an agile, tool-using robot-astronaut that can outlast any human on a space walk. SCOUT, a lunar rover being developed by NASA, will carry astronauts but will also have the potential to act on its own.
Partners in life
Last month, NASA launched two competitions to encourage the private sector to create autonomous robots ' ones that can assemble structures with minimal human intervention and ones that can steer along a flight path and touch down to take surface samples.
The urge to make robots more human isnt just about sending them to space. Robots will eventually wheel around hospitals, schools, and homes ' and they will need to be able to read social cues, learn a task, and be able to do work without supervision. The latest version of Honda Motor Co.s ASIMO robot debuted last month in Japan. The four-foot-tall robot can push a cart and run like a person. Such a human-like robot would be like having an assistant, a sous chef in a kitchen who might not be as good as the chef, who might not be able to do as much work as the people, said former astronaut Hoffman. With robotic assistance, people will be much more useful, and, conversely, by having people around, robots will be much more useful. (NYTNS)