Lockheed, Rand, Schlumberger, Texas Instruments, and Xerox-PARC. Other institutions in this country have shown increasing interest in the field. International activity is concentrated in Great Britain, Japan, and the Soviet Union, with some work under way in Canada, France, Italy, West Germany, and Sweden.
These research and development programs are necessary for the eventual success of the applications described elsewhere in this report. They are also a part of the environment which has led to NASA's current strong interest in the potential of machine intelligence in space. However, even a vigorous research effort does not necessarily imply an applications development process adaptable to future NASA needs. The technology transfer problem is further aggravated by the relative scarcity of qualified workers in the AI field. NASA may begin to alleviate this manpower crisis by directly supporting artificial intelligence and robotics research in colleges and universities throughout the United States.
1.2 History of NASA Automation Activities
Since its inception in the late 1950s, NASA has been primarily devoted to the acquisition and communication of information about the Earth, the planets, the stars, and the Universe. To this end, it has launched an impressive string of spectacularly successful exploration missions including the manned Mercury, Gemini, and Apollo vehicles and the unmanned Surveyor, Mariner, Pioneer, Viking, and Voyager spacecraft to the Moon and beyond. Numerous Earth orbiting NASA satellites have added to an immense, growing fund of useful knowledge about terrestrial resources, weather and climatic patterns, global cartography, and the oceans. Each mission has made use of some level of automation or machine intelligence.
Mission complexity has increased enormously as instrumentation and scientific objectives have become more sophisticated and have led to new problems. The Mariner 4 mission to Mars in 1965 returned about 106 bits of information and was considered a tremendous success. When Viking revisited the planet only a decade later, roughly 1010; bits were returned with greatly increased flexibility in data acquisition. Even now, the amount of data made available by NASA missions is more than scientists can easily sift through in times on the order of a decade or less. The situation can only become more intractable as mission sophistication continues to increase in the future, if traditional data acquisition and handling techniques are retained.
A 1978 JPL study suggested that NASA could save from $500 million to $5 billion per annum by the year 2000 AD if the technology of machine intelligence is vigorously researched, developed, and implemented in future space missions. According to a special NASA Study Group:
"Because of the enormous current and expected advances in machine intelligence and computer science, it seems possible that NASA could achieve orders-of-magnitude improvement in mission effectiveness at reduced cost by the 1990s [and] that the efficiency of NASA activities in bits of information per dollar and in new data-acquisition opportunities would be very high" (Sagan, 1980). Modern computer systems, appropriately programmed, should be capable of extracting relevant useful data and returning only the output desired, thus permitting faster analysis more responsive to user needs.
During the next two decades there is little doubt that NASA will shift its major focus from exploration to an increased emphasis on utilization of the space environment, including public service and industrial activities. Current NASA planning for this eventuality envisions the construction of large orbital energy collection and transmission facilities and space stations operated either in Earth or lunar orbit or on the surface of the Moon. The first steps toward space industrialization already have been taken by NASA's Skylab astronauts who in 1973 performed a number of successful material-processing experiments. Progress will resume when the Space Shuttle delivers the first Spacelab pallet into orbit, and this line of experimentation continues.
Economy is perhaps the most important reason- why robotic devices and teleoperated machines will play a decisive role in space industrialization. A conservative estimate of the cost of safely maintaining a human crew in orbit, including launch and recovery, is approximately $2 million per year per person (Heer, 1979). Since previous NASA mission data indicate that astronauts can perform only 1 or 2 hr of zero-gravity extravehicular activities (EVA) per day, the cost per astronaut is on the order of $10,000/hr as compared to about $10-100/hr for ground based workers. This suggests that in the near term there is a tremendous premium attached to keeping human beings on the ground or in control centers in orbit, and in sending teleoperated machines or robots (which are expected to require less-costly maintenance) into space physically to perform most of the materials-handling jobs required for space industrialization.
In summary, the objective of NASA's space automation program is to enable affordable missions to fully explore and utilize space. The near-term technology emphasis at OAST includes:
- Increasing operational productivity
- Reducing cost of energy
- Reducing cost of information
- Enabling affordable growth in system scale
- Enabling more cost-effective high performance missions (planetary, etc.)
- Reducing cost of space transportation
