NASA to test info tech in deep space

As part of an upcoming mission into the distant reaches of the solar system, NASA will shift its focus from carrying out science experiments to testing cutting-edge technology as it sets the stage for future space exploration.

The unmanned Deep Space 1, the first mission in NASA's New Millennium Program, will provide a test bed for applications in the areas of navigation, on-board automated operations and spacecraft systems monitoring. Each form of technology is designed to reduce cost and mission risk while at the same time reducing the spacecraft's dependence on human ground control.

Wesley Huntress, NASA's associate administrator for space science, characterized the $152 million mission as the space agency's first "true technology mission." The experimental technologies are crucial for future missions, he said.

"The technological advances that we're proving here...are those that can make or break many of the missions that we have planned," Huntress said.

Deep Space 1 is due to be launched in late October or early November from Kennedy Space Center, Cape Canaveral, Fla. One mission of the spacecraft is to gather information on celestial bodies. During the next year, Deep Space 1 is expected to cross the path of an asteroid, and, possibly years later, the craft will fly by a distant comet.

One of the most critical IT experiments on board Deep Space 1 will test a new autonomous navigation system, called AutoNav.

In traditional solar system missions, ground controllers track a radio signal from a spacecraft to determine its position. However, Deep Space 1 will dramatically stray from this traditional navigation process by allowing the spacecraft to take over parts of the navigation job formerly carried out by ground controllers.

The spacecraft will find its location in the solar system by taking images of known asteroids with an on-board camera and comparing the asteroids' positions to background stars. Using positions of 250 asteroids and 250,000 stars stored in the memory of an on-board computer, AutoNav can determine the craft's position and make any corrections necessary to stay on course.

The most crucial test for AutoNav will come in July 1999 when it will attempt to guide the spacecraft within three miles of a two- to three-mile-wide asteroid at 35,000 miles per hour. This will be the closest flyby of a solar system body ever attempted. Based on optical images collected, AutoNav will plan the spacecraft's flight path and order any correction maneuvers needed.

"[The spacecraft] actually will be closer to the asteroid than you are to the surface of the Earth when you fly on a commercial airline," said Marc Rayman, Deep Space 1's chief mission engineer and deputy mission manager. "We have confidence that the system will be accurate enough so that we don't hit it."

NASA also hopes to reduce dependence on ground controllers for many spacecraft functions using a new artificial-intelligence program the agency developed called Remote Agent.

Artificial-intelligence software exhibits human intelligence and behavior, including the ability to learn and adapt to change based on past experience.

With Deep Space 1, instead of assigning specific tasks to the software program as they have done in the past, ground controllers assign generalized tasks and allow the software to "decide" how to act.

The software package generates a list of required activities and feeds the list to an "executive" component that decides how to carry out the activities by taking into account variables such as the state of the spacecraft, constraints on operations and the high-level goals provided by the ground controllers. This executive component also monitors responses to these commands and modifies them if the response was not what was planned.

The Remote Agent will generate and execute plans on board the spacecraft to command spacecraft altitude, on-board camera imaging and thrusting of the propulsion system. The software also will be fed simulated command-response failures, such as the on-board camera failing to be turned off when it was commanded to do so. While taking into account other factors, such as how much power the camera will consume by continuing to run, the software will modify operations to account for the camera's failure to respond to the command.

Douglas Bernard, the Remote Agent experiment project element manager at NASA's Jet Propulsion Laboratory, said the software package is designed to allow future spacecraft missions to obtain scientific data while cutting costs due to a decreasing reliance on ground controllers.

"We want the spacecraft to be smarter," Bernard said. "We want the operation of the spacecraft...to have a remote agent. It acts as an agent for the operations in terms of doing some tasks [controllers] would have to do themselves."

Finally, the mission will be testing a new way for the spacecraft to relay information about its general health or condition to ground controllers.

During the normal operations of the spacecraft, various telemetry data— such as on-board temperature and pressure measurements— are gathered. However, this data accumulates, and sometimes it is not possible for ground controllers to download all the data during the one eight-hour pass dictated by the Earth's rotation.

Deep Space 1 will test new artificial-intelligence software that allows the spacecraft to summarize data pertaining to its health— highlighting the most important information— so that controllers can quickly respond to any "urgent" tones that indicate the spacecraft has a serious problem, said Jay Wyatt, group supervisor for Deep Space 1's advanced multiple mission software technology group. This method also will reduce the time needed to downlink the data.

"You are capturing the most important stuff. For mission risk, that's important," Wyatt said. "It's enough information for [controllers] to do trouble-shooting."

John Logsdon, director of the Space Policy Institute at George Washington University, Washington, D.C., said this mission represents a major shift in NASA's approach to space exploration by "investing in technology and testing it before you build a mission around it.

"Space is a harsh environment," Logsdon said. "It should reduce the risk of future missions; you don't have to fly untried technology."