- Swarm robotics explained
- Why more (bots) is more
- Swarm robotics applications
In nature we see creatures like insects, birds, and fish collaborating as a unified system to achieve a common goal, leading to quicker and more accurate results than what would be achieved if each individual would work on its own. Scientists have now taken this collective intelligence and applied it to robots, which can offer a wide variety of benefits and help us solve some of our most pressing real-world issues. Potential applications include assistance in environmental matters or search and rescue operations, but as swarm intelligence improves and enables even more autonomous capabilities and enhanced collective behaviours, the possibilities are virtually endless.
Swarm robotics explained
Much like natural systems, such as in swarms of insects, schools of fish, or flocks of birds – swarm robotics systems consist of many individual but interconnected parts with the aim to collectively solve problems by applying swarm behaviours like organisation, navigation, decision making, and so on. Natural swarms collaborate and communicate without any central control and (re)act and adapt according to changes in their immediate environment. This enables them to pursue common goals and perform complex tasks. Natural swarms are extremely adaptable, scaleable, and robust, and can quickly disperse or aggregate in the environment. Their pattern formation abilities enable swarms to assemble in various shapes in order to carry out a certain task. While the individuals of natural swarms generally have limited abilities and don’t usually possess general knowledge of the task at hand or the swarm they’re in, research has shown that very intelligent and complex group behaviour can, however, emerge through communication among swarm members and the resulting transmission of information.
While swarm robotics is still in its infancy, various research platforms are focussing on transitioning swarm robotics solutions from theory to prototype industrial systems and applications.
These natural swarm abilities, insights, and properties are also used to create robots that are simple in behaviour and structure, but in a swarm can interact and collaborate. Interaction among swarmbots can either take place directly (robot-to-robot) or indirectly (robot-to-environment), and the collective behaviour that results from these interactions enables the ‘swarm system’ to carry out complex tasks – just like the swarms in our natural environment. And while swarm robotics is still in its infancy – with current systems relying on centralised communication and control infrastructures – various research platforms across the world are working on transitioning swarm robotics solutions from theory to prototype industrial systems and applications.
Why more (bots) is more
So while the individual parts of a robot swarm are relatively simple and hardly impressive, in collaboration with each other they can carry out some pretty mind-blowing tasks that might not be easy or even possible for traditional, standalone robots to accomplish on their own. The swarm approach, therefore, has many advantages.
Scalability
Scalability is one of the most important advantages of swarm robotics. The ‘rules of engagement’ can remain the same, irrespective of the size of the swarm. And there is no need to keep all the robots in a swarm connected to a central station or command centre, as the interactions are predominantly local and achieved through ultrasonic sensor technology and transmission systems like radio frequency.
Flexibility
Their reliance on interacting with the local environment means that swarms are very flexible and adaptable and can quickly respond to changes. Swarm bots are able to find solutions by collaborating with each other and changing their roles according to the task at hand, or act simultaneously according to changes in their environment. A couple of small changes in the algorithm controlling the individual bots in a swarm can result in an enormous variety of complex swarm behaviour, with bots completing other bots’ tasks and collectively adapting to whatever needs to be done.
Robustness
As the members of the swarm are all equal, swarms are enormously robust. A swarm of small, cheap robots, each with limited capabilities, can be a good alternative for one larger, highly capable and expensive robot. And one swarmbot perishing – or environmental disturbances or system faults – will have little to no effect on the operational ability of the swarm. Furthermore, the control of the botswarm is decentralised, which means that there is no single robot (or human) controlling the operations of the others.
Swarm robotics could potentially be used to create artificial bees and insects to pollinate crops to ensure their survival and prevent ecological disasters.
Swarm robotics applications
Even though swarm robotics is still relatively young and has not yet been widely adopted across sectors, the development of a number of applications is already well underway. Currently, swarm robots are mainly used in military and exploratory operations, but they could soon also be implemented in industries like mining, agriculture, and so on. In scenarios like fire-fighting operations, a swarm of unmanned aerial vehicles (UAVs) could support firefighters by gathering and relaying critical information on factors like changing weather and fire conditions. Swarm robotics is also very useful for environmental monitoring, searching for survivors during natural disasters, or cleaning up oil spills. Robot swarms can be deployed for collective exploration, establishing communication networks, optimising interconnected transportation networks, acquiring situational overviews, searching for objects, and monitoring the environment. Swarm robotics could also potentially be used to create artificial bees and insects to pollinate crops and other important plants to ensure their survival and prevent ecological disasters. Inconspicuous robot swarms could also work as real-time collectors of information that is critical for the performance of all kinds of interconnected systems.
Here are some real world examples of swarm robotics in action.
“This is the first time there’s a swarm of drones successfully flying outside in an unstructured environment, in the wild. In the next few years we will be able to have very reliable systems.”
Enrica Soria, Swiss Federal Institute of Technology Lausanne
Swarmbots autonomously fly through dense bamboo forest
Recently, during an impressive experiment by scientists at Zhejiang University, a swarm of ten purpose-built, palm-sized drones with on-board computers, altitude sensors, and depth cameras, managed to autonomously navigate a dense Chinese bamboo forest. The swarm avoided branches, dodged embankments, and passed through the narrow spaces between bamboo stalks, while finding the most optimal route through the woods. What really sets this drone swarm apart is its ability to determine its environment on its own, map it and then plan its trajectory, using algorithms for optimal flight efficiency, collision avoidance, and coordination within the swarm. The scientists conducted several experiments with the swarm, including a flight in which the drones were forced to stay in formation and one in which the swarmbots needed to fly criss-cross and demonstrate their ability to avoid colliding into each other or into a moving person. Xin Zhou, who led the experiment, said: “Our work was inspired by birds, which fly smoothly in flocks, even through dense forests.” And Enrica Soria, a roboticist at the Swiss Federal Institute of Technology Lausanne, says: “This is the first time there’s a swarm of drones successfully flying outside in an unstructured environment, in the wild. In the next few years we will be able to have very reliable systems.” As these drone swarms do not make use of GPS or other external infrastructure, they could be deployed for tasks like natural disaster relief, such as during earthquakes or floods, to determine where help is needed or assess damage.
Swarm vehicles make warehouse logistics more efficient
The future of warehouse logistics has arrived. The LoadRunner prototype transport vehicles, developed by the Fraunhofer Institute for Material Flow & Logistics in collaboration with warehouse equipment and automation technology supplier KION, operate in a swarm – using distributed artificial intelligence – and can reach speeds of up to ten metres per second. The perfectly coordinated LoadRunners are ideal for sorting tasks in warehouses, and they can collaborate by magnetically linking to each other in order to enable the transportation of large objects. A parcel sorting test operation yielded impressive results; in one hour, just 60 LoadRunners managed to sort more than 10,000 shipments, assisted by AI for traffic management. The vehicles use 5G to communicate with each other, can independently accept orders, and autonomously coordinate their routes. What’s more, if one LoadRunner malfunctions, the distribution centre only needs to replace that particular one, while the others in the swarm carry on with their work.
AI-controlled drone swarm autonomously locates Hamas terrorists
In May last year, in what appeared to be a world’s first – the Israel Defense Forces (IDF) managed to locate, identify and attack Hamas terrorists using an AI-controlled drone swarm. Only a single human operator was required to initially direct the swarm, after which the drones – as a connected unit – guided themselves. The swarm consisted of small aircraft communicating with each other to locate the targets and direct the airstrikes, which paints a grim picture of the recent advances in AI-driven combat, and reaffirms existing concerns over autonomous robots being used in the military. Arthur Holland of the United Nations Institute for Disarmament Research, said: “They are certainly a notch up in the incremental growth of autonomy and machine-to-machine collaboration in warfare”. According to an IDF commander, the swarm unit has already carried out more than 30 operations and plans are underway to make more drone swarms available to support the IDF.
Swarm robotics’ traits and capabilities can offer significant environmental and economic benefits to smart agriculture initiatives across the world.
Adaptive swarm robotics could assist in smart agriculture tasks
Researchers from Texas A&M University System are working on the development of a configurable, scalable, and adaptive swarm system that consists of unmanned aerial vehicles (UAVs) and ground robots. The swarm system enables waste reduction, optimal use of fertiliser, and minimal use of water. The relatively small machines in the swarm also reduce soil compaction, and their application of non-chemical weed control methods helps reduce herbicide-resistant weeds. All of these traits and capabilities offer significant environmental and economic benefits to smart agriculture initiatives across the world. “We will develop the technical and theoretical groundwork for the deployable, scalable swarm system consisting of a physical robotic swarm, of both ground and aerial robots, a digital twin simulator for low- and high-fidelity simulations, and an easy-to-use user interface for farmers to make this CASS system into use,” says Kiju Lee, one of the project leads. Lee continues: “Current trends in precision agriculture and smart farming mostly focus on larger machinery or a single or a small number of robots equipped and programmed to perform highly specialised tasks. This project will serve as a critical pathway toward our long-term goal of establishing a deployable easy-to-use swarm robotic system that can serve as a universal platform for broad agriculture applications.”
In closing
Robot swarms need to be able to operate in the real world, adapting to dynamic changes, and coping with events and external conditions that are difficult or even impossible to model or predict. The field of swarm robotics has seen significant advances in recent years, and capabilities and traits like collective decision making, autonomy, scalability, and adaptability, make swarm robotics extremely suitable for tackling real-world challenges. Potential applications include search and rescue operations, precision agriculture, logistics, surveillance, and many others. According to Airlie Chapman from the Melbourne School of Engineering, “UAVs will become more inconspicuous, acting as big data collectors. Smaller vehicles will be a core component here, quietly collecting real-time, bird’s eye information critical for interconnected systems to perform well in aggregate.”
Indeed, the potential for swarm robotics to provide solutions for many of our challenges appears virtually unlimited, which could make it one of this century’s most significant technological advancements.
