ASA is not just about opening air and space frontiers; we're also about building the tools we need to open those frontiers. While we've had some great successes, there's much more to do. We want to cut our system costs by about an order of magnitude. We want to cut our cycle time of development by a factor of 3 to 5, so we can build and launch a spacecraft within a year to a year and a half. We want to improve reliability by up to a factor of 10,000. To achieve these goals, we need bold, revolutionary leaps - not just at NASA, but in engineering and engineering education.
NASA has a bold strategic vision for the next 25 years. We want to develop technologies that cut travel time across the Pacific in half and cut costs and improve safety so space travel is not limited to our astronauts. We want to be able to make annual to multi-decade predictions of climate, environment and resource management to promote sustainable development on this planet. Within 10 to 15 years, we want to be able to detect Earth-sized planets around stars up to 100 light years away and see if those planets support life. Within 25 years, we want to be able to see geological and biological processes on those distant bodies. And we want to use the International Space Station as a platform so an astronaut can leave Earth orbit and maybe one day visit Mars or work on a research station on a near-Earth asteroid.
To make our vision a reality, we will need more intelligent systems, more flexible modular vehicles, breakthroughs in miniaturization, better, lighter materials, and advanced operating capability. For example, to send a probe to a distant star, where it will be too far away to receive operational commands from mission control, we'll need a spacecraft that can think for itself and learn and adapt as it goes. It will have to be self-diagnostic and self-repairing; in many ways, it will have to be like the human body. It will have sensors and actuators, it will react to stimuli, and it will adjust to changing environments.
Traditional numerical approaches will not work in such a spacecraft. Instead, we'll need soft computing that more closely resembles human intelligence - we need a vehicle IQ. We will develop modules from the chip level through the system level. With a modular design, we can assemble similar modules into different spacecraft to achieve different missions. This building block approach will also find its way into vehicles operating closer to Earth. For example, modular design will enable us to modify wing configuration to change the role of an aircraft.
To go beyond the solar system, we'll also need smaller and cheaper spacecraft. We've already made significant strides in that area. Viking, which landed on Mars in 1976, cost over $3 billion (in today's dollars), was about the size of a small car, and took about a decade to develop. By contrast, Mars Pathfinder cost less than $300 million, was much smaller and took less than three years to develop.
In the future, the costs and the sizes of spacecraft systems will continue to shrink as their capabilities grow. In the near term, we plan to build spacecraft about the size of a television set. Ultimately, we'll develop nanospacecraft that will weigh less than a kilogram, fit in the palm of your hand and have their entire avionics on a single chip. These spacecraft, which will operate in some of the harshest environments known, will also benefit from advances in materials and design tools.
It's not just the spacecraft that must get smarter and smaller. We must also develop intelligent, streamlined control systems and continue to reduce the work force needed for space missions. For every person you see in the space shuttle mission control, there are many others behind the scenes. Launching the shuttle takes thousands of people and a few hundreds of millions of dollars. NASA's new X-33 experimental vehicle, which will fly next summer, will require no more than 50 people for operations. The Viking program required about 1,000 people in mission control; Pathfinder took about 50. Future missions will require 12 or fewer.
How do we get from here to there? How do we take the engineering design culture from where it is today to where it must be for the missions of tomorrow? For a long time, engineering was a pencil-to-paper culture. In the 1960s, we moved from slide rules to electronic drafting boards and wire-frame computer modeling. By the mid-1970s, we were using solid models to represent geometry and three-dimensional surface contours. Despite these advances, too much time was still wasted by the incompatibility of the analytical models used in such disciplines as aerodynamics, thermodynamics, structures and controls. The design process was sequential, with separate discipline groups. System design was optimized at the discipline level, not the system level. Data and design information had to be moved from one group to another - a task accomplished by people carrying large piles of paper.
About 20 years ago, we merged design with manufacturing. CAD/CAM, which significantly reduced design cycle, process time and engineering change orders, led to concurrent engineering - linking diverse disciplines through the use of digital data sets. The best example of concurrent engineering is the development of the Boeing 777. At the peak of design work, 238 design teams comprising 6,000 engineers, using data from 4,000 worldwide computer terminals, manipulated three trillion bytes of information representing 20,000 design releases.
But we still have a disconnect from discipline to discipline. We still have many distributed, unconnected databases across engineering fields and manufacturing. And because we don't have the capacity for real-time simulation, we can't get a real-time visualization and integration of all these tools. NASA is working to break this logjam. In our Product Design Center at the Jet Propulsion Laboratory, we are bringing disciplines together, which has enabled us to reduce analysis of preliminary mission design concepts from half a year to two weeks.
Because of the sequential nature and limitations of our tools, there is still far too much uncertainty throughout the life cycle of a product, there are too many people involved, and we have just begun to address the geographically distributed nature of what we do. We need to capture design knowledge earlier in the design process and learn to deal with the unprecedented quantity of data that confronts us and convert it into usable knowledge.
Because of current shortcomings, we commit about 90 percent of the cost of a new product when we only have about 10 percent of the knowledge about it. And the more we incur costs in any design process, the more our flexibility to make changes, without the risk of cost and schedule overruns, diminishes. As a result, we don't get an optimized design, products cost too much, and they're not as effective as they should be. We must eliminate the discrete steps of conceptual design, preliminary design and final design, as well as manufacturing training, maintenance and operations.
How do we close the gap between design knowledge and cost commitments? At NASA we're using something called Intelligent Synthesis Environment (ISE). We're not just updating tools, we're fundamentally changing the culture of engineering. The major components of ISE are human-centered computing, infrastructure for distributed collaboration, rapid synthesis and simulation tools, life-cycle integration and validation, and the cultural and educational change we need to inject into the creative process.
Human-centered computing deals with the dynamics and interfaces between the human being and the computer. Our current way of interfacing with the computer is for WIMPS (Windows, Icons, Menus, Pointing Systems). It's not the way we deal with our real environment. We interact with the world, process information and make decisions based on all of our senses. In addition, there simply isn't enough coupling of humans and computers. We need to exploit research on human biology and the brain's cognitive processes. Perhaps someday we'll talk to our computers using alpha waves or the electrical activity in our muscles.
We must also learn to discover and integrate various sources of data, using active intelligent agents to "mine" data and present it to the user in a variety of formats, including text, numbers and multimedia images. This will help us move from data, to information, to knowledge, to intelligence.
Next, we need to build the infrastructure for distributed collaboration so we can take full advantage of diverse teams around the world. We need at least a hundred- to a thousand-fold increase in computer capability to achieve the ISE vision. We need to move into nonsilicon or nonelectric computers - perhaps to optical and biological computing.
That's why we also need to increase our networking capability. The amount of information flowing through the pipeline must increase from under a gigabit per second, where it is today, to at least one hundred to one thousand gigabits per second. Tomorrow's networks must have intelligent switches and routers, or intelligent interfaces. The increased networking ability will enable us to link computers, mass storage facilities - and people - seamlessly.
Though separated geographically, scientists and engineers will be able to work in collaborative teams in the engineering design process. They will be able to work together, using the full range of human senses, even if some of them happen to be on Mars. Each team member will be able to participate and communicate from his or her own creative perspective. They will be free from the keyboard and the terminal, which are literally locking up our creativity.
The third part of ISE is rapid synthesis and simulation tools. Today, because of limitations in our models, we oversimplify the real world. We rely on separate complementary test programs to establish worst-case operating and failure conditions to guide our analyses. To account for uncertainty and to quantify risk, we must move from traditional deterministic methods to nondeterministic methods, like probabilistic approaches, neural networks, genetic algorithms and symbolic computing.
We've already achieved a high level of sophistication in numerical simulations across many disciplines, but we need to do more. Today's technology limits us to about one billion nodal connections and one billion nodal interactions per second. We need to aim for the capability of the brain, which is more than one million times more powerful. Only rapid analysis and optimization capacity near this level will close the design knowledge-cost commitment gap and enable us to go from mission requirements, to multidisciplinary analysis and design, to simulation of manufacturing and virtual prototyping, to operations and repair, to product disposal. The virtual design process will also give us, with unprecedented detail, cost impacts and risk-level assessment.
To ensure that we have analytical models to verify real-world behavior and failure mechanisms, we need to integrate analytical models development in real time with experimental testing. This approach will dramatically shorten the cycle time of product development by enabling a seamless flow from initial concept through final design and manufacturing.
To date, industry has concentrated on simulation of manufacturing, planning and processes. We have simulators of individual machines and we have real-time assessment of inventory flow control. But we need to be able to simulate an entire factory before we erect it; an entire plane before we build it; an entire mission to Mars before we launch it. Then we can begin to simulate operations, including repairs and maintenance.
That brings us to the next component of ISE: integration and validation. At this point, there are many unknowns. We don't know what fidelity we need. We don't know what scale is required. We don't know how collaborative virtual teams will work. These are fundamental issues. To address them and demonstrate and validate the future collaborative design environment, NASA plans to establish national, virtual distributed testbeds - geographically distributed computing environments that integrate hardware-in-the-loop, real-time information operating systems, and all associated engineering design tools.
At NASA, we want to focus these testbeds in critical areas such as reusable launch vehicles, next generation space telescopes, the space station as an engineering center, human exploration, and environmental modeling. To do so, we'll need broad industry and academic involvement. That's a cultural barrier, because this is not just about the aerospace industry. It is not just about technology. It's about people and how people work and communicate on a global scale to enrich our lives. Perhaps most of all, it's about education.
Back when we were using slide rules, people referred to the Three R's. Now we need the Three I's: information, integration and infrastructure.
Information: Engineers need more of it, or at least more of the right kind. Today's engineers can use the "black box," but they didn't build it. Because they didn't build it, they have trouble understanding the limiting first principle assumption of how it works, which leads to design and application failures. It's not that we have bad people; our teachers and students are the best. We're just guilty of using old approaches in a new era.
Integration: We need to do more to integrate the most rapidly advancing fields in engineering education, such as biology. We are on the cusp of the Biological Revolution, and it's leaving engineering in its wake. We've seen how the knowledge explosion in the biological sciences, particularly in areas like health care and agriculture, has changed the way we live. The engineering community can't afford to sleep through this revolution.
We must take the intellectual underpinning of biology and apply it to engineering. Some call this biomimetics. I've referred to the power of the human brain. A single brain cell can carry 10 times as much information as today's most powerful computer and it is much more energy efficient. The engineering sciences need to apply this capability. We must understand, appreciate, integrate and exploit biological processes and their incredible potential. Our goals at NASA depend on that, as does the future of engineering. This is something our education system must reflect.
Infrastructure: Government, universities and industry must realize that our educational infrastructure can no longer be bricks and mortar; it must be bits and bytes. This means we must initiate and support revolutionary changes in our university engineering programs and create new capability for lifelong learning.
NASA and government must continue to push the envelope with long-term, high-risk research. We must work with industry to translate research concepts into workable methods and products. Most of all, we must bring in universities as fundamental, intellectual participants in the development of technology crucial for the ISE vision. NASA is already planning a program to do just that.
Industry must also support not just operations and near-term product development, but long-term, high-risk research and development. We'd like industry to work with us to create the workforce of the future, one that is not only more flexible, but that recognizes creativity and has a deeper understanding of the systems and tools it develops and uses.
And universities must understand this changing environment and develop new curricula and research capabilities. This is absolutely crucial because it is tomorrow's engineers who will enable everything I've discussed in this essay. With his invention of the liquid-fueled rocket, Robert H. Goddard, a 1908 WPI graduate, launched the Space Age. Future Goddards may pilot a submarine beneath the icy ocean we think covers Europa. Others may build a colony on Mars. They will have the training, the understanding and the tools. But only if we work together. So let's get to work.
- Goldin is administrator of the National Aeronautics and Space Administration (NASA). This article is adapted from Goldin's keynote address to members of the Industry/University/Government Roundtable on Enhancing Engineering Education, who met at WPI in May.
Last Updated: 11/19/98 22:35:10 EST