Swarm Intelligence Explored | Are two (or more) heads actually better than one?

February 28, 2020

Swarm Intelligence Explored | Are two (or more) heads actually better than one?

Since the dawn of time, nature has been man’s muse. A quick Google search would reveal as much, providing tons of information which supports the fact that mother nature has been the inspiration for numerous scientific breakthroughs and inventions throughout history. 2020 Commemorates the birth of a new decade, but even in the nub of the tech revolution, we still look towards nature for ingenuitive solutions. Swarm Intelligence is one such solution, proving that seemingly unintelligent creatures can achieve amazing things when working together. 

Swarm intelligence (SI) in nature can be observed in the collective behaviour of social colonies and focuses on the notion that simple creatures following simple rules can achieve highly complex and intelligible results. Ants, for example, can only complete simple tasks in isolation, but an ant colony exhibits behaviour indicative of intelligent planning, such as building bridges and enslaving other ant species. Numerous algorithms and metaheuristics have been developed based on SI, and the concept is even utilized to develop biologically-inspired AI Systems. Applications of these systems are prevalent in numerous industries, fueling remarkable innovations utilised in Space Exploration, the Military and Environmental Conservation, to name a few. 

SI Military Applications

There are numerous examples of SI-based military weapons and systems, but a particularly interesting application involves Navy vessels. In 2014, the US Navy demonstrated the use of autonomous boats on the James River in Virginia. The demonstration consisted of 13 unmanned patrol boats, equipped with sensors, escorting a “high value unit”. The sensors gather information about their surroundings and all information is continuously shared between the boats. This “hive-sense” network of sorts allows the autonomous boats to operate in sync whilst escorting assets, choosing routes and interdicting enemy vessels. The Office of Naval Research (ONR) spearheaded this project as a way to save lives and cut costs. 

Other areas of the globe have also realised SI’s potential for future warfare purposes. Flock-93, a concept proposed by Russian researchers and engineers, involves swarm drones armed with explosives, capable of identifying and swarming enemy targets. These swarms would in theory consist of over 100 drones operating autonomously, whilst sharing information with one another. This visualisation of the future of warfare has led to DARPA (U.S. Defense Advanced Research Projects Agency) investing over $316 000 in an SI study conducted by Buffalo University. This study consists of utilising gamers’ brainwaves to teach swarms of autonomous robots (both ground and air based) how to strategize and work together to complete missions. 

OffWorld's Robots (Image: Offworld)

SI in Space Exploration

To quote Jim Keravala, “There’s a bleeding edge of our society which yearns for the frontier”. Keravala is the CEO of OffWorld, a California based startup which develops industrial robots that can be used both on earth and in space. OffWorld’s robots rely on Machine Learning (ML) and are powered by onboard solar panels. OffWorld is currently building robots with a vast variety of abilities, such as harvesting ice for drinking water, paving surfaces and establishing power plants, to name a few. They envision a future where robots work together using swarm intelligence to “prepare” planets in outer space for human habitation. Their robots weigh around 53kg and are able to operate in a wide variety of extraterrestrial environments, such as asteroids, the Moon and even Mars. Although OffWorld’s vision may seem a bit far fetched, they have landed support from numerous Fortune 500 companies which intend to use their technology for resource mining in space. 

The World's First Walking Rover (Image: Spacebit)

Spacebit, a UK based company, is also using swarm intelligence to develop legged robots intended for moon exploration. These spider-like robots are able to explore crevices, cracks and lunar caves which traditional rovers are unable to explore, and will utilise AI to gather in depth information which will hopefully shed light on the history of the moon. The mission would involve Astrobotic Peregrine Lunar Lander Spacecrafts operating as motherships and transporting numerous legged robots to the dropoff-point, at which point these legged robots will leave the spacecrafts and explore the moon in a swarm. Spacebit’s robots weigh around 1.3 kg each and can be built in 6 to 12 months, rapidly decreasing the standard spacecraft construction schedules which normally spans over years. According to Pavlo Tanasyuk, Spacebit’s CEO, the first test mission will fly in 2021, making Spacebit the first company to send a swarm of legged robots to space. 

SI Environmental Applications

In arid regions, floods and heavy rainfall leads to erosion and the loss of precious soil. In an effort to reduce flood related erosion, Roboticists at Harvard’s Wyss Institute developed Romu, a robot that can autonomously drive interlocking steel sheet piles into the ground, The robot leverages its own weight to apply downward force whilst driving sheets into soil, interlocking the sheets and ultimately forming a continuous wall, providing proper soil stabilisation. Justin Werfel (Senior Research Scientist) states that Romu robots should optimally work in a swarm, responding in real-time to changes in environmental conditions and adapting to situations quickly and collectively whilst still operating autonomously. Romu swarms could potentially also be useful in situations where human labor resources are scarce or when heavy machinery struggle to access sites. 

Romu In Action (Image: Wyss Institute at Harvard University)

In addition to developing Romu, Researchers at the Wyss Institute are also developing RoboBees which can potentially be used in crop pollination as well as high-resolution weather, climate and environmental monitoring. Inspired by numerous flying insects, these independant aerial systems are smaller than a paperclip and are capable of coordinated behaviour in a large group. RoboBees are able to complete numerous tasks autonomously such as; flying and hovering, dynamically responding to their environment and working within a RoboBee colony to coordinate activities and behave as an effective unit whilst completing tasks. 

Where to next?

Systems equipped to function autonomously and respond to their physical environment are bound to bring about transformative change. One can only speculate what the future may hold, but one question remains: If animals and insects alike can amplify their intelligence through swarming, could humans achieve the same?

Sharen van Lill

Marketing Manager

Swipe iX Newsletter

Subscribe to our email newsletter for useful tips and valuable resources, sent out monthly.