What ants, termites and bees can teach us about traffic control
What ants, termites and bees can teach us about traffic control
America’s love affair with the automobile has been glamorized by the image of the lone driver in a sports car on an empty, country road. The pavement glistens as the morning sun strikes the freshly dewed surface … and, well, we all know the rest. We know the reality of our current highway system as well: the tragedy of accidents, lost time in traffic jams, and the degraded environment and health.
Transportation is now the second most expensive consumer item in the United States, behind housing, and takes up nearly 18 percent of the average household budget. Moreover, the 13 million gallons of oil consumed every day, at a cost of about $1 billion, are largely wasted, according to the Rocky Mountain Institute. Less than 0.5 percent of the energy in the fuel of the typical automobile actually moves the driver. The rest is busy heating up the road and tires and moving the mass of metal that is the car.
Many innovations are being developed to address these issues: alternative fuels, lighter and stronger vehicle bodies, new drivetrains and recharging systems. One solution set that has not gotten as much attention, but could yield real short-term dividends, has been inspired directly by nature.
It’s called swarm intelligence, and is the phenomenon that you might see in a flock of birds or school of fish or swarm of bees. Despite twists and turns and dives, the collection of animals seems to move as one, and no one individual collides with another. The same mechanisms at work in the swarm could someday eliminate the accidents and lost time that currently cost the United States nearly $400 billion a year.
What makes a flock of birds stick together
In the flock of birds, for example, three activities were discovered to be taking place and have been modeled: alignment, separation and cohesion. Each bird steers in the average heading of the flock, stays a certain distance away from his neighbors, and steers according to the average position of these neighbors.
This is an example of so-called “emergent behavior.” Here the eventual outcome is not dictated by one rational choice, but by the direct and indirect interactions of individuals acting alone, without a global awareness. High-level or complex behaviors are produced by the interaction of these individuals who are performing simple acts. As you might imagine, this is describing a system: It has parts, relationships and a collective outcome.
Indeed, this phenomenon cannot be defined below the system level -- meaning it can't be predicted by reducing it to its constituent parts, just as you cannot define your mind by a cataloging of neurons.
In the animal world the eusocial insects (ants, bees, wasps and termites) are the masters of this behavior. When ants forage, they lay down pheromone trails as they bring back food. Each searching individual then follows the strongest (most traveled) trail, and found food is efficiently carted back to the nest. Termites exhibit this behavior when they build their impressive mounds, bees when they build their beautiful hives.
Photo of peak hour traffic provided by Steve Lovegrove via Shutterstock.
How stock trading is like a traffic jam
In the world of technology we can see similar system behavior -- often called multiple agent systems or MAS -- in the fluctuations of the stock market or the formation of traffic jams. We have made use of these principles in a wide variety of applications to move information, in routing data packets in telecommunications, to move material in designing delivery or repair routes, and to move energy in optimizing power grids. Internet searches, customer preferences, and market predictions have all been enhanced by the use of these “bottom up” simulation models.
Southwest Airlines has used an ant foraging model to design a gate assignment system. Each plane "remembers" the gates with the shortest delays and queues accordingly.
Regen Energy of Toronto manufactures an autonomous wireless power controller for the household that minimizes peak electricity demand and communicates with other units to do so. Icosystem of Cambridge, Massachusetts has organized field crews for more efficient gas and oil drilling using an "evolutionary computation" based on insect foraging.
Our current highway system is a basic example of the kind of system in which these agent-based algorithm principles could be applied. Changing the timing and spacing of cars on the road, without even touching their power systems, could result in a 33 percent improvement in efficiency for urban drivers, according to estimates. That's better than the 20 percent gain we could get from converting gasoline-powered cars to hybrids. A relatively cheap sensor system would be employed to create telematics-enabled vehicles.
Inching toward the driverless car
Many research and development groups are working on combining look-ahead capabilities with on-board reaction mechanisms. Here developing a more sophisticated adaptive cruise control (ACC) with real time local conditions is essential.
The three components of the approach are: the ACC itself, implemented in the form of a car-following model; an algorithm for automatic real-time detection of traffic; and a strategy matrix to adapt driving characteristics to traffic conditions. This means integrating sensors in the infrastructure and in the vehicle.
Another approach concentrates on the communication between vehicles and objects.
Volvo has vowed to produce an injury-proof car by 2020, and engineers there have studied the flight habits of locusts to gain insights in how to do it. Locusts fly in the millions without colliding -- and they do so because their neural circuitry sends visual input directly to their wings for near-instantaneous flight adjustment. Volvo is working on a “sensory input routing methodology” that mimics this.
The company is also currently working on its Safe Road Trains for the Environment (SARTRE) project. This initiative is a semi-autonomous “follow the leader” convoy technology and is a half-step between current automobiles and autonomous vehicles. It completed its first public road test in 2012.
Cameras, radar and laser technology present in current safety systems have been extended to enable a car to keep pace precisely with a lead vehicle driven by a professional driver. The road test involved one truck followed by three cars that were driven autonomously at 56 miles per hour and spaced as close as 13 feet.
This system allows for the driver to exit the train at will and for following vehicles to catch up and fill the gap left. It also reconciles the conflict between efficient long haul transportation and the need for independent movement in the last mile of a journey. Volvo estimates the system could be employed on a typical highway within ten years.
Nissan Motor Co. has also been active in the research and development of autonomous vehicles for both safety and congestion relief. In 2009 it unveiled its Eporo robot car concept based on the schooling behavior of fish.
Like the flocking model, collision avoidance was based on relative positioning, but in the Nissan model behavior algorithms were defined by concentric zones. The robots were programmed to avoid collisions in the near zone, travel side by side, matching distance to speed, in the next zone, and to catch up to any unit in a far zone. The company has used these concepts in refining its “Safety Shield” technology, a six-stage collision avoidance system.
Making cars behave like fish
Nissan collaborated with the government of Beijing in a 2012 pilot project to reduce traffic congestion by providing drivers with timely information. Twelve thousand cars were equipped with portable navigation devices connected to a central traffic information center via cell phone technology. The onboard devices received real-time conditions and showed drivers the fastest routes to their destinations, thus saving time and fuel.
In a further step toward autonomous vehicles, Nissan in October announced an Autonomous Emergency Steering System designed to avoid rear-enders and sudden intrusions in low-speed zones. The system employs front mounted radar and camera, and rear radar and laser scanners, to first chime a warning, then show an escape pathway on a display and finally to take over the steering of the car if a clear path is available.
These advancements are part of a trend toward smarter, more efficient vehicles and systems that has, currently, many disparate streams to it. It will be interesting to see how some of these improvements will be combined.
Will lighter vehicles with an alternative drive train be combined with intelligent traffic control systems or hybrid mass transit? Will a disruptive technology obviate the need for some of the current designs? Will land use and cultural change shift the desires of the driving public? These and many more questions remain in the exciting world of transportation design.
Whatever the changes ahead I believe nature will continue to inspire some of our solutions. Here we have seen that rearranging vehicles in time and space alone can yield great energy savings. As in nature, just-in-time information , which I have written about before, is essential. Finally, emergent behavior, like swarm intelligence, can be used to design simple, adaptable and robust solutions to many of our traffic problems.