0
Select Articles

# The World is More Than ComplicatedPUBLIC ACCESS

Complex Systems of the Future will have to be Adaptable, and New Approaches will be Needed to Engineer Them.

[+] Author Notes

Ahmed K. Noor is Eminent Scholar and William E. Lobeck Professor of Modeling, Simulation, and Visualization Engineering at Old Dominion University in Norfolk, Va.

Mechanical Engineering 133(11), 30-35 (Nov 01, 2011) (6 pages) doi:10.1115/1.2011-NOV-1

## Abstract

This article discusses the need of complex systems to be adaptive, and various innovative technologies that are required to engineer these systems. Complex adaptive systems consist of several simultaneously interacting parts or components, which are expected to function in an uncertain, complex environment, and to adapt to unforeseeable contingencies. The defining characteristics of complex adaptive systems are that the components are continually changing, the systems involve many interactions among components, and configurations cannot be fully determined in advance. Studies have shown that complex systems of the future will require a multidisciplinary framework—an approach that has been called emergent (complexity) engineering. Emergent engineering designs a system from the bottom-up by designing the individual components and their interactions that can lead to a desired global response. Although significant effort has been devoted to understanding complexity in natural and engineered systems, the research into complex adaptive systems is fragmented and is largely focused on specific examples. In order to accelerate the development of future diverse complex systems, there is a profound need for developing the new multidisciplinary framework of emergent engineering, along with associated systematic approaches, and generally valid methods and tools for high-fidelity simulations of the collective emergent behavior of these systems.

## Article

Technology is always pushing limits. What is impossible today becomes commonplace tomorrow. Think of the telephone in your house when you were a child. If you are an adult today, you’ll remember that you couldn’t take it out of your pocket and ask it about your flight schedule and airport terminal, send text messages, videotape an event with it, or use it to translate a Japanese e-mail into English.

In a little more than a century, aerospace technology progressed from glider experiments to robotic explorers orbiting the planets and moons of the solar system, and exploration of the surface of Mars using largely autonomous solarpowered rovers.

Technology tackles ever-greater challenges and so almost always increases the complexity of fielded systems over the systems of previous generations. Over the years, technologists have kept pace by developing tools and facilities to manage engineering processes.

Swarm on Mars: Modular robots that can reorganize themselves into different configurations could adapt to changes in terrain.

These have generally been top-down design approaches, based on systems engineering, decomposing the design into separable components, characterizing the intended relations among them, and verifying that the system is built and operated as intended.

There is mounting evidence, however, that current methods of systems engineering are reaching their limits. The Airbus A380 aircraft, which entered service in 2007, was two years late and over budget by more than 2 billion euros. The delivery of the first Boeing 787 Dreamliner is expected in the fourth quarter of 2011, more than three years late with a cost overrun of more than $10 billion. NASA's James Webb Space Telescope is estimated to cost$1.2 billion to $1.8 billion more than the$5 billion estimate of 2006. The launch date is expected to be at least seven years later than the date set in 2004.

Traditional engineering approaches seem to be failing with the A380, the Dreamliner, and the Webb telescope because the systems have large numbers of parts and extensive software (millions of lines of embedded code), and each of the designed systems is only a component of a larger system, involving distributed design, acquisition, and manufacturing teams, and a variety of processes. A large number of interactions and changes made by individual teams during the development process can result in unexpected consequences.

The sum of parts count and source lines of code has been used as a possible metric for the complexity of a system in the Defense Advanced Research Project Agency's Adaptive Vehicle Make portfolio of programs, initiated in the summer of 2010. The AVM portfolio aims at reducing the time of delivery of future military vehicles, including aircraft.

Cutting-edge engineering projects today involve a range of disparate technologies and disciplines, which must come together to create a cohesive whole. They can be heterogeneous mixes of electronics, sensors, actuators, embedded software, mechanical linkages, and motors. The systems are designed as networks of components. Interaction among components is nonlinear and can give rise to unpredictable responses. In addition, several distributed teams participate in virtual enterprises to create a product.

Researchers expect that the systems of tomorrow will be more complex than those of today and so will need to have structures that are adaptable to deal with the unforeseeable. In short, they will be complex adaptive systems.

## Complicated vs. Complex

The complicated systems that engineers dealt with for many years consisted of decomposable functional components, or subsystems, which are well defined, and which interact with each other in well-understood ways. The systems usually operated as intended throughout their useful lifetime.

Consider a microprocessor with over 3 billion transistors. It is indeed a complicated system, which must be carefully designed, with every transistor located in its place with utmost precision. The tasks to be performed, the operating environment, and all contingencies are considered in the design, and nothing, as far as possible, is left unspecified. The same was true for considerably less-complicated microprocessors. The Intel 4004, introduced in 1971 had 2,250 transistors. But with 2,250 or 3 billion transistors, the major difference between the microprocessors is scale. The same well-understood design principles apply for both devices.

The Web site, created as a companion to Mechanical Engineering magazine's feature focus, contains material on complex adaptive systems, cyber-physical systems, emergent (complexity) engineering, along with the classes of complex adaptive systems (and systems-of-systems), swarm and modular robotics, smart power grids, emergency response systems, and smart cities. There are also links to other online services and features of the Center for Advanced Engineering Environments at Old Dominion University.

There are also two related Mechanical Engineering articles published earlier this year:

The development of a practical, modular and reconfigurable robot is the subject of the feature “Smart and Modular” in the September 2011 issue; research conducted at the University of California at Berkeley into cooperating UAVs was the subject of an article, “Airborne, Autonomous, and Collaborative,” in the April 2011 issue.

A complex system, on the other hand, involves uncertainty. Complex adaptive systems consist of several simultaneously interacting parts or components, which are expected to function in an uncertain, complex environment, and to adapt to unforeseeable contingencies.

A case in point would be a swarm of autonomous modular robots, a system design that is being considered for future exploration of Mars. Concepts have been proposed by researchers at NASA, the European Space Agency, the Massachusetts Institute of Technology, the Institute of Process Control and Robotics in Karlsruhe, Germany, and the University of Southampton in the United Kingdom.

A swarm of robots searching for methane on the surface of Mars must have the flexibility to operate in an unknown environment, collaborate in performing tasks, and adapt to unforeseen situations. As generally conceived, the robots are made up of modules that reassemble themselves to meet the requirements of different tasks and environments. Centralized control and predetermined script execution are not practical. The swarm must have a high degree of redundancy to allow for fault-tolerance, and to compensate for the failure of some of the individual robots.

The robots need miniature cameras, sensors, actuators, and computational elements to observe their environment and to communicate and collaborate with neighboring robots. They need fuel cells to keep their electronics and sensors operable. From their local observations, individual robots must make autonomous decisions and take action. Emergent behavior of the swarm, including its new configurations, result from the individual actions of the robots.

Testing waters: A current research project seeks to launch a fleet of robots able to monitor underwater environments.

The defining characteristics of complex adaptive systems are that the components are continually changing, the systems involve many interactions among components, and configurations cannot be fully determined in advance.

The individual robots of the swarm can start a local exploration in the form of a four-legged structure, but when they encounter a small cave, for instance, they can reconfigure their modules to assume a snake-like structure and crawl through the space.

The swarm involves interactions among components on different levels—the modules that make up the robots, as well as the robots themselves—and control is highly decentralized. Because there are many, if some modules or robots malfunction, others remain to carry on.

Some of the interactions are unanticipated because of the nature of the local terrain. The configuration of the robots cannot be programmed beforehand, but rather must change to meet changing needs.

These are the principles behind such familiar complex adaptive systems as ant colonies, flocks of birds, social networks, the Internet, and the stock market.

## Emergent (Complexity) Engineering

The overall goal of traditional engineering approaches is to develop a stable, reliable, controllable system, like a microprocessor, with predictable performance to carry out pre-defined tasks in a bounded environment. This is accomplished through the steps used in the development process— functional specification, design, testing and validation, and then manufacturing.

Complex systems of the future will require a multidisciplinary framework—an approach that has been called emergent (complexity) engineering.

It is a convergence of complexity theory and science with such disciplines as communications, sensor technology, mechanics, computational intelligence, and control theory. The goal of emergent engineering should be to produce robust complex systems, operating in uncertain environments, and capable of adaptation and change.

“Emergence” describes a phenomenon that occurs at macroscopic scales, but not at the microscopic. An example from classical mechanics is the friction between bodies that emerges when their surfaces rub against each other. It is a non-conservative force, even though the forces between the elementary particles are conservative.

Emergent engineering designs a system from the bottom-up by designing the individual components and their interactions that can lead to a desired global response. Each robot consists of modules that communicate and interact with neighboring modules, and the swarm consists of robots that communicate and interact to perform as a group. The design enables individual robots and the entire swarm to change in response to issues that emerge at the macroscopic scale. Control is distributed, and if one part fails, the others continue.

Another advantage of the system is that it can be altered to take on additional tasks by introducing a few more robots, without a need to redesign the entire swarm.

In Search of the Swarm

The advantages of sending complex adaptive systems, specifically robotic swarms, where it is difficult or dangerous for humans to go—from ocean depths to extraterrestrial planet surfaces—has spawned a number of active research programs. Some initiatives are directed towards novel applications of swarm robotics and cooperation between humans and the swarms.

So far only prototypes have been developed with a limited number of robots capable of communicating with each other in real time. The full potential of the swarm concept, with perhaps thousands of robots, is yet to be realized.

The Scripps Institution of Oceanography of the University of California, San Diego, with funding from the National Science Foundation, is building a fleet of miniature autonomous underwater explorers to gather information on localized currents, temperatures, salinity, pressure, and biological processes.

Initially, a small fleet of 20 1.5-liter robots will be used. Each robot is the size of a soccer ball, and has a complete set of electronics for controlling its buoyancy. The robot also has temperature and depth sensors, a hydrophone for recording subsea sounds, an inexpensive inertial navigation sensor, and a fiber optic encoder for measuring the shaft displacement of the linear actuator, used for controlling and adjusting buoyancy. A GPS unit and satellite communication module are used for communicating the robot's location, when it is at the surface.

Researchers at the Swiss Federal Institute of Technology in Lausanne have developed a swarm of small and lightweight flying robots, each equipped with a module in the wing to emit a wireless signal and enable communication among rescuers in disaster areas. Each robot can locate rescuers and establish a communication network with a base station.

A research team from the Free University of Brussels, funded by the European Commission, has been working on the development of humanoid robotic swarms in what it calls the Swarmanoid project. A group of heterogeneous, dynamically connected small autonomous robots capable of moving in 3-D space, the swarm can organize and distribute a given task into subtasks to be performed by different groups of robots. This is accomplished by having each group of robots specialized in certain activities, like climbing, ground transport, and so on. The Swarmanoid the team developed had 60 robots divided into three groups, “eye bots” for sensing and analyzing the environment, “foot bots” for moving on rough terrain and transporting either objects or other robots, and “hand bots” for climbing and grabbing objects.

The Micro Autonomous Systems and Technology (MAST) project being developed by Georgia Institute of Technology in collaboration with the University of Pennsylvania, the California Institute of Technology, and NASA's Jet Propulsion Laboratory uses vision-based and navigation technologies to create swarms that can gather information from areas too dangerous for the humans to enter. The swarm can survey and quickly build a detailed floor plan map of an entire structure, and beam it to nearby responders. The next step will be to add aerial robots to the team, which can locate particular buildings and potential entry points. The work is sponsored by the U.S. Army Research Laboratory.

Researchers at the Swiss Institute of Technology in Zurich have developed the concept of a distributed flight array, a multi-propeller system consisting of individual robots that dock with each other and fly as a swarm. The robots exchange information and combine this with their sensor measurements to determine the thrust needed for take-off. The system can be useful for carrying larger loads than can be handled by individual robots.

The Flyfire project of Massachusetts Institute of Technology intends to create a swarm of automated miniature helicopters with embedded color-controlled light emitting diodes to act as smart pixels in a massive 3-D display as they fly through the air. The pixels can change color, and their precisely controlled 3-D dynamic movement can create an immersive experience from any angle.

The swarm can transform itself from one shape to another, or morph a twodimensional photographic image into an articulated shape. Each of the individual micro-copters of the swarm has a mass of 13 grams, a wing span of 23 cm, and is equipped with two motors for the thrust and lift needed for its movements. Only a few micro-copters have been deployed, but plans are to significantly increase the number. Potential new applications include advanced public displays, and dynamic forms of telepresence.

The U.S. Navy has initiated a program to develop a swarm of micro-robotic machines that are capable of manufacturing complex objects, possibly other robots. The micro-robot assembly is expected to be capable of manipulating nano- and micro-scale building blocks and performing basic operations such as pick and place, removing material, and joining components.

While human-scale and other large robots have been developed that can perform impressive tasks, there are a number of applications for which swarms of tiny cooperating robots are capable of doing much more than a single large robot. For example, a swarm of small autonomous collaborating unmanned aerial vehicles can be used for more efficient surveillance over a wide area than can be done by a single large UAV.

Also, the speed and range of the regional and local exploration of planets can be improved by using swarms of autonomous flying and surface robots, instead of a large single flying or surface robot. In addition to wide coverage, the swarm improves fault tolerance, enables expandability, and allows for parallelization of simple tasks.

## A Look at the Future

Although significant effort has been devoted to understanding complexity in natural and engineered systems, the research into complex adaptive systems is fragmented and is largely focused on specific examples. Diverse future applications have been identified, covering many different scales. These include fine-grained systems (e.g., swarms of nano satellites), intelligent and autonomous vehicles, and very large-scale complex adaptive systems of systems, such as cyberspace. The principles may some day be applied to create smart cities, in which services ranging from transportation and energy to education, health care, and public safety are integrated and optimized.

Disperse and explore: Many small robots cover more territory than a single large one. If some fail, the mission can continue.

To accelerate the development of future diverse complex systems, there is a profound need for developing the new multidisciplinary framework of emergent engineering, along with associated systematic approaches, and generally valid methods and tools for high-fidelity simulations of the collective emergent behavior of these systems. The framework and tools should be tested on diverse systems at a variety of scales.

Emergent engineering is a 21st century metaphor relating innovation, complexity, and cybernetics, within a new intelligent environment in which people will be surrounded by intelligent tools and communicating devices, which are sensitive and responsive to them and are seamlessly integrated in the environment. The environment will be filled with sensor webs (sensors smaller than the eye could see, joined together into networks larger than the mind could comprehend), wearable electronics, ultra intelligent electronic agents, telepresence robots, and computers that respond to brain waves. The purpose will be to protect and to serve, and not only to make humans more productive, but also to support them in the enjoyment of their lives.

View article in PDF format.

## Errata

Some tools below are only available to our subscribers or users with an online account.

### Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections