newslife site

Ebi maste cheshat

Ebi maste cheshat remix

Add comment 2009

The Mysterious Field of Engineering Systems

One of the nation’s revered technology leaders dispenses anecdotes and wisdom on the slippery subject of engineering systems (or systems engineering). Norm Augustine just can’t get a handle on the discipline: “No one agrees on what it is, or what it does.” After years in industries like Lockheed Martin, Augustine has come up with “Norm’s Rules,” and can at least define ‘system’ as “having two or more elements that interact,” and ‘engineering’ as “creating the means for performing useful functions.” But these definitions don’t get you too far in the real world.

Augustine shows a fuel control system, which some engineers might view as part of a propulsion system. In turn, aeronautical engineers might think of the entire airplane as a system, and transport engineers view aircraft as merely components in systems incorporating airports, highways, shipping lanes. Augustine continues up the ladder until “our system that started as a fuel controller…seems to have the whole universe as a system.” Like Russian Matryoshka dolls, systems can always be embedded within larger systems. Even if you try to simplify a system in terms of just a few objects with a binary, on-off interaction, things can get complex very quickly. Five elements in a system can exist in more than a million possible states. Says Augustine, “A typical earth satellite has nearly one million parts; a 747 over 5 million. How does that make you feel about flying?”

Distinguishing the significant interactions and the important external influences on a system are central to designing and problem solving. And these days, engineers must include politics, public policy and economics as part of their systems. “The trick is to bound the scope of the system so it’s not too large to be analyzed and not too small to be representative.” Doing this right is “why systems engineers should be paid so much.”

Augustine concludes with his “Dirty Dozen” systems engineering traps, which have led to embarrassing bust-ups, monumental failures, and real tragedies. Notable among these: “the ubiquitous interface,” (or absence thereof). He describes how two flight control groups used different metric units and accidentally sent a Mars-bound spacecraft whizzing off into deep space. There’s the “single-point failure,” exemplified by the collapse of a football field-sized satellite dish due to a poorly designed bracket. There’s software, “which like entropy, always increases:” a Mariner spacecraft headed in the wrong direction due to a missing hyphen in 100 thousand lines of code. The problem with most systems ultimately is that they “contain human elements … and humans sometimes do irrational things.”

Add comment 2009

MIT Perspective on Engineering Systems

The field of systems engineering has only recently emerged, and as this symposium demonstrates, defies precise definition. But MIT has taken this evolving area to heart, nurturing a new division and encouraging a raft of ventures that in their execution, may help shape the field for the next century.

An MIT freshman in 1900 had some very specific requirements to fulfill for graduation, and to prepare for a responsible role in society, says Subra Suresh. Courses included mechanical drawing, military science and rhetoric. These choices became richer over time, with the addition of hundreds of engineering faculty, dealing increasingly with the sciences. Suresh traces how over many decades an engineering concentration on metallurgy shifted from studying mining (iron), to aviation (aluminum), plastics, electronic materials and then biological materials. But at each step, he notes, MIT “always lagged behind about 10 years” in what it taught students.”

The Engineering Systems Division (ESD) is an attempt to “train people the right way.” The curriculum brings the basic rules of nature into engineering practice, and applies discoveries to products and processes that impact people. Students must take into account the “long term societal impact.” ESD is needed to link complex issues along technological and social dimensions. The modern engineer must create new ideas and technologies, and reinvent tools and technologies from earlier times — as Suresh puts it, “Fix problems associated with the greatest achievements of the 20th century.”

Yossi Sheffi fine tunes the picture, enumerating the key domains under the ESD umbrella, as well as the approaches faculty have adopted, in research, teaching and real-world projects. The primary distinction between other engineers and ESD engineers, Sheffi notes, is that “we try to look at the big picture.” So ESD focuses on critical infrastructure (water, transportation), such extended enterprise as supply chain management and global factories; energy sustainability and health care delivery. To get a handle on such unwieldy subjects, professors examine the human-technological interface, and delve into uncertainty, dynamics, design and implementation, networks and flows, and policy and standards.

MIT’s “engineers without labs” are seeking to “develop insights, principles and tools across all systems,” forging partnerships in industry, around the world. ESD students and faculty must get out in the field, says Sheffi, not just to fulfill course requirements but in order to tackle significant global problems, and to find solutions that are sustainable in terms of social equity, economic development and environmental impact. ESD values and accepts “intellectual risk,” meaning issues that may appear unquantifiable or vague, even without solution, and understands that problem solving means respecting and bringing together all disciplines, including the social sciences and management.

Add comment 2009

Looking Ahead to 2020

Real-world practitioners of systems engineering/engineering systems describe how the young discipline has shaped their very large enterprises.

For the past 10 years, David Lehman has been incorporating key systems engineering ideas within MITRE Corporation. Successes include getting project leaders to think about engineering solutions in the context of political and economic organization, and learning how to communicate these solutions better. MITRE has talked to defense acquisition managers in the field to extract data and create models that get disseminated to other managers. But Lehman is disappointed that Defense Department acquisition methods are still large-scale, and unresponsive to swiftly changing situations. He’d like to show program managers how “to step outside what they’ve been taught,” and create incentives for doing the right things rather than “sticking with regulations.”

Robert Skinner, Jr. wonders if engineering systems approaches can help with some pressing questions: the way to mix transportation and land use decisions in urban areas, for instance, or government pricing strategies for surface transport. One nettlesome issue involves the right scope of analysis, says Skinner. Should researchers be looking at the components of the transportation system, or the whole enterprise? “As we move downward, uncertainty increases and the role of social systems and social science enters into it; politics upper and lower case becomes more significant.” And he adds, “We’re sorely lacking in analogs in the policy world to transmit complex engineering concepts. If analysis gets too far out ahead of the public’s and decision-makers’ ability to absorb it, it all comes to naught.”

“Why are so many complex systems behind schedule and over budget?” asks Heinz Stoewer. A single line of code missing can cause system collapse, says Stoewer. And big problems can flow from human shortcomings in calculations, accounting or risk management. Stoewer believes another reason for failure is that program managers and systems engineers “are too process focused,” and not well enough aligned. They may lack sufficient depth in the key discipline of their projects, leading to faulty product design or production. To improve the chances of success, Stoewer emphasizes the importance of early phases: “I can tell you two dozen programs in trouble because they’re…making enormous efforts trying to get things right when they’re almost done.”

By 2020, Joel Moses hopes that engineering systems will be recognized “as having made significant contributions” to health care, energy, environment, financial services and the military. To achieve such an impact, the field should focus on “maybe the key issue” of system architecture. Each engineering field thinks of architecture in different ways and groups must communicate better with each other. Moses believes educators should teach “what makes for a good system architect,” and that “systems thinking is important, but not enough.” A good system architect sees things holistically. Moses notes as well, “the difference between designing a one-off versus a family of systems.”

Add comment 2009

How the Brain Encodes Reward

As Ann Graybiel puts it, “basal ganglia were dark basement structures” until Okihide Hikosaka began his classic 1980s research demonstrating how these neuronal clusters influenced eye movements. Hikosaka has deepened and broadened his work in this once neglected area of the brain, and brings a McGovern audience up to date on his latest discoveries.

Hikosaka briefly sketches what is known about the basic pathways leading in, around and out of the basal ganglia, circuits that have been associated with stress, pain, mood, memory and arousal. This specialized cluster of neurons seems especially attuned to the neurotransmitter dopamine, and Hikosaka has been investigating “a number of unsolved questions,” including how dopamine neurons form circuits for movement control, whether such neurons encode “motivational values,” and what other parts of the brain guide them.

Hikosaka describes research demonstrating that certain dopamine neurons become excited if a visual cue indicates a future reward, and become inhibited with a visual cue indicating no reward. Dopamine also increases after an action delivers a reward and decreases when an action produces no reward. Research began to explore whether dopamine neurons “encode motivational values, including reward and punishment.” After others’ studies yielded contradictory or uncertain conclusions, Hikosaka designed a set of studies on monkeys involving classical Pavlovian conditioning, with juice rewards and air puffs as aversive stimuli.

Among Hikosaka’s findings: some dopamine neurons were excited primarily by positive, reward-predicting stimuli, others inhibited by air puff-predicting stimuli. But he also found another group of dopamine neurons excited both by positive and negative reward-predicting stimuli (as well as the stimuli themselves). Hikosaka posited two types of neurons that react in very different ways to motivational signals, which he described as value-coding and salience-coding. He also determined that the lateral habenula, a part of the brain sitting at one end of the thalamus, seems to regulate dopamine pathways involved in some motivational responses. By sending a weak electric pulse through the lateral habenula, Hikosaka saw a very strong inhibition of the dopamine neurons that “encode mostly motivational values.”

Add comment 2009