Computers, Use of

views updated

Computers, Use of

The earliest human spaceflights were guided by navigational computers on the ground; there was no onboard computation. But starting with project Gemini , computers have been an essential part of every space mission. When the first piloted Gemini flew in 1965, most computers were the size of a room, and so it was a remarkable technological achievement to shrink a computer down to a size (2 cubic feet) that could fit into the small capsule. Onboard computing power enabled Gemini to carry out tasks such as rendezvous and docking even though the computer was underpowered by today's standards. It contained 4,000 words of memory, about a thousandth the size of a handheld personal digital assistant today. The Apollo computer that controlled the lunar landing had only 32,000 words of memory.

Early Spaceflight Computers

The computers used in spaceflight have always been a mixture of leading and lagging technology. The fast chips used in desktop and laptop computers on Earth would never survive in space because cosmic gamma ray radiation would deposit electrical charges on the chips, and cause data loss or other failures. Therefore, many chips used in space are custom-designed with redundant circuitry: three circuits instead of one so that the three can vote on the correct answer and ignore a single incorrect result caused by cosmic radiation . In other cases standard chips can be protected from radiation with special metal shielding, but even then the onboard chips are typically ten or twenty times slower than Earth models.

The experience of the National Aeronautics and Space Administration (NASA) on the Apollo program changed the way people thought of software as a component in a large system and ultimately led to great advances in the software development process. In 1966 NASA was concerned that the software might not be ready by the scheduled launch of Apollo 1. Until that time software had been thought of as a minor add-on to large projects. Now it appeared that software development delays were threatening the space race with the Soviet Union. NASA and its partner, the Massachusetts Institute of Technology (MIT), were forced to develop better practices of software requirements analysis, documentation, verification, and scheduling. Eventually they were successful, and many of the practices they developed remain in effect. The Software Engineering Laboratory at NASA's Goddard Space Flight Center is still a leader in the field.

The Uses of Computers in Spaceflight

Computers are used in spaceflight for three purposes: to reduce costs, reduce risks, and increase capability. The most significant form of cost reduction lies in minimizing ground operations. For example, scientists at NASA's Ames Research Center developed an artificial intelligence program to automate scheduling of the space shuttle ground processing, a task with roughly 10,000 steps. The program saved time and money, and led to the spinoff of a successful company that provides software to constantly monitor manufacturing variables, such as customer demand and resource availability, thereby helping Fortune 500 companies optimize their factory operations.

The speed and reliability of computers have enabled complex space missions and maneuvers such as bringing the space shuttle back from orbit to take place with a reduced risk of failure. However, computers also play an important role in risk reduction before a mission is even launched. During the design stage, computer simulations search for problems and computerized failure analysis techniques estimate the probability of failure and point out areas to improve.

Computers enable human spaceflight but also diminish the need for it. When Wernher von Braun first imagined space travel, he thought that an orbiting space station would be staffed by about eighty scientists observing the weather and performing other tasks. He did not foresee that unmanned robotic satellites would perform most of those tasks more efficiently and less expensively. Astronauts are so expensive that robots are preferred wherever possible, and are relied on exclusively for all exploration beyond low Earth orbit and the Moon.

There are two kinds of robotic control: telerobotic and autonomous. In telerobotic control a human guides the movements of a robot in another location via radio signals. A fascinating example is Robonaut, a human-sized robot with two arms and hands, a head, a torso, and one leg. Under development at NASA's Johnson Space Center, Robonaut is designed to carry out space walks under the control of a human in the safe environment of the space station or on the ground. Robonaut has hundreds of sensors, giving the human operator a feeling of actually "being there."

Autonomous control is used when a telerobotic link would be too slow or too expensive to maintain. For example, Mars is typically about twenty minutes away from Earth by radio communication, and so rovers on the Martian surface are designed to have some autonomous control over their own actions. For more ambitious missions, such as the Mars sample return mission currently scheduled for 2014, more capable autonomy using artificial intelligence will be required. Autonomous robots are also useful as assistants to humans. An example is the Personal Satellite Assistant, a softball-sized robot designed to float in the weightless environment of the space station. It is designed to propel itself by using ducted fans, take pictures, analyze temperature and gas levels, and communicate by voice control. It can check on the status of the station and assist astronauts in doing experiments, using a combination of autonomous and telerobotic control.

The best uses of computers combine the three attributes of cutting costs, reducing risks, and increasing capability. An example is the Remote Agent program, which controlled the experimental Deep Space 1 mission in 1999. Using technology similar to the space shuttle's ground processing scheduler, Remote Agent allows ground controllers to send a high-level command such as "take pictures of this star cluster" rather than detailed low-level commands such as "open valves 3A and 4B, then burn the engine for three seconds." The program comes up with the best plan for achieving the high-level goal and then executes the plan, all the while checking to see whether something goes wrong, and if it does, figuring out how to fix it.

NASA administrator Daniel Goldin has stated: "When people think of space, they think of rocket plumes and the space shuttle, but the future of space is information technology." Advanced computer technology will continue to contribute to this future.

see also Humans versus Robots (volume 3); Robotic Exploration of Space (volume 2); Robotics Technology (volume 2); Simulation (volume 3); Telepresence (volume 4).

Peter Norvig

Bibliography

Heppenheimer, T. A. Countdown: A History of Space Flight. New York: John Wiley &Sons, 1997.

Muscettola, Nicola, P. Pandurang Nayak, Barney Pell, and Brian C. Williams. "Remote Agent: To Boldly Go Where No AI System Has Before." Artificial Intelligence 103 (1998):5-47.