At a symposium concerned with the advances and regulatory issues pertaining to the burgeoning area of Unmanned Aerial Systems (UAS, aka "drones"), Dr. Richard Kelley of the University of Nevada at Reno brought up something that is still oscillating in my head: In almost every article or discussion on drones, the machines are talked about in the singular, i.e., the drone. What he brought up is so logical an extension of Moore's Law that it seems positively inevitable. We're going to see, everywhere and in abundance, drone swarms.
That's right, swarms as in things that flock together
Similar to the most impressive of nature's swarming creatures, insects, birds and fish, drone swarms will consist of compact individual units that gather together to efficiently perform tasks. When a task is completed, the swarm can instantly deconstruct and its units, or autonomous elements, can be rapidly redeployed onto other tasks, possibly involving different groupings for optional efficiency.
Another hot area that we tend to speak about in the singular is the Internet of Things. The vast majority of discussion surrounds the sensing or input side. Much less of the conversation is about sensor data coming into IT systems, and the output is in the form of analytics and visualizations.
However, the converse of sensing is actually controlling. Robots, for example, require all three (sensing, analytics and control) in order to function. Now bring these two ideas together, drones in the plural and IoT vis-à-vis sensing, analytics and control functions. Now you can begin to imagine swarms of what are, in effect, flying or driving robots.
What makes a swarm?
At the most basic level, swarms are based on murmuration, or flocking. For objects in motion to flock, there are three primary rules that must be applied:
1. Avoid collisions with adjacent "flock mates"
2. Match velocity with adjacent "flock mates"
3. Stay as close as possible to adjacent "flock mates"
Beyond these, if flock mates have different characteristics, via a master command and control mechanism, the appropriate flock mate can be assigned to a specific swarm and task set.
Why should we care?
First of all, they're coming. Much like smartphones and WiFi, they're going to start appearing everywhere. This is not an exaggeration -- early work on robotic swarming is taking place at universities and companies already, and there's no reason to think that the ultra-fast pace of technology adoption that we're experiencing is slowing. So, expect the swarms to arrive.
Secondly, they can offer great benefits in the context of both critical operations and society as a whole, as with other mass-adopted innovations. Swarms will start to appear in the sky and on the ground, possibly in our homes. Some will consist of insect-sized flying robots, others that essentially represent the evolution of the vacuuming robot. Most likely, the incarnations of robotic swarms, much like how smartphone apps have evolved, will be of types and functions that we cannot yet imagine.
Business cases for swarming machines
The first areas where swarming is already starting to appear are primarily in transportation. Witness the demonstrations by the Freightliner division of Daimler Trucks involving big rigs that can perform "platooning," which is essentially a linear form of flocking. In Europe, a large project is underway to develop "Safe Road Trains" which can take the platooning concept even further. On the world's oceans, autonomous swarming vessels, currently under development by the U.S. Navy, can be imagined as extending into container ship operation and other efforts to increase safety and efficiency. Manufacturing and warehousing logistical operations can potentially be greatly enhanced with the advent of small, specialized swarming robots.
To swarm or not swarm?
In nature, biologists have empirically determined that in most formations, depending upon the creatures being studied, there is no leader of a swarm. This is incredibly fascinating, but in the context of robots and business applications, some amount of remote or centralized control is usually desirable. Other applications will be viable with each unit in a swarm making decisions entirely based upon its onboard sensors, software and what it learns from flock mates around it. Whether swarming makes sense for a given application will likely depend upon the cost of implementation and operation, the capabilities of the robotic units and what they can accomplish when constituted as a swarm, versus employing a more traditional method. As an example, some large scale farms are already considering swarming for seed-spreading and harvesting.
This is one of those big disruptors
It's not outrageous to suggest that at some point we'll look out of our windows and see aerial swarms working nearby, or rolling swarms performing tasks on the street. Doubtless due to the operational and cost efficiencies that swarms can potentially offer, they will in various forms start to appear in commercial, industrial and governmental sectors, and possibly even our residential environments, in coming years.
Science fiction has provided a glimpse into large swarms, and they're coming at us in, well, droves.
This article is published as part of the IDG Contributor Network. Want to Join?