The edge of Swarm Robotics

Published in

IET-VIT

6 min readFeb 25, 2021

Image depicting a swarm of minibots, from bistrolroboticslab.com

A swarm is defined as a large group of beings, usually associated with flying insects which move together to perform essential tasks collectively. Individually, each unit of a swarm is essentially not that functionally equipped, but as a swarm, they can work complex tasks. Swarm robotics naturally follows the same features as that of a swarm. Each robot itself isn’t as efficient at a particular task as it would otherwise be, in a swarm.

Observably almost all of robotics is based off of designs available in nature itself. Actually, all of modern technology is a way to replicate what nature has to offer for us. Similarly, swarm robotics is a field of multi-robotics in which a large number of robots are coordinated to work in a distributed and decentralized way. It is based on the use of local rules and simple robotics, and is mostly inspired by the nature of some swarms, such as social insects, fishes or mammals, which interact with each other in the swarm in real life.

The behavior of the swarm itself is constructed through the careful study of existing animals functioning in swarms, such as migration of bird crowds, fish shoals, foraging of ants, bee colonies and so on. It solely depends on the intelligent group behavior exhibited by individuals with poor abilities.

Goals and important Constraints:

Since the quantity of the robots in a swarm is vital in this field, it is important to focus on making the group activities intelligent and the individuals as simple and cost efficient as possible.

The design of the swarms is guided by the swarm intelligence principles. The system should have the ability of realization of fault tolerance, scalability and flexibility.

Now, in all social insects, the individuals are not informed about the colony’s overall status, or, the global community. There exists no leader that guides all the other individuals in order to accomplish their goals. All the knowledge of the swarm is distributed through its constituent agents, and a situation is created where an individual is not able to accomplish its task without the rest of the swarm. This is the behavior expected of our swarm robots.

Stigmergy and Swarm intelligence:

Stigmergy is a mechanism of indirect coordination, through the environment, between agents or actions. Every individual of the swarm interacts with another and the interactions are based on the locality. These interactions are passed on through the colony and therefore the swarm can solve tasks that could not be solved by a sole individual.

Self-organization relies on the combination of the following four basic rules: positive feedback, negative feedback, randomness, and multiple interactions.

This collective behavior is defined as self-organizing behavior and the robots in a swarm can mimic this mechanism of stigmergy through swarm intelligence. Swarm intelligence deals with the discipline of decentralized control and self organization in artificial systems.

Properties of a Swarm intelligent system:

The following properties define most Swarm intelligent

The individuals of the swarm must be autonomous and simple. There should also be a significant number of these robots so as to be able to control a complex system.
Individuals are mostly all homogenous, or composed of different types among similar topologies.
Interactions between the individuals are solely based on simple communication between them directly, or through the environment around them, their decisions of organization depending only on the surrounding environment. There is decentralized control.

Image depicting the property of self organization of a swarm, from idsia

Technicalities:

For designing the swarm, the applications must be at the collective level, but the hardware must be at an individual level since that defines behavior. The designs can be divided into manual design and automatic design.

In manual design, mostly, a trial and error process is followed where the designer of the swarm develops and improves the structure after multiple tests until the desired collective behavior is achieved. The most commonly adopted software architecture in swarm robotics is the probabilistic finite state machine which incorporates advanced methods such as computational linguistics, machine learning, time series analysis, speech recognition, machine translation, circuit testing, and many more, for pattern recognition. This mechanism helps create most of the aggregate collective behaviors of the swarm.

Another way of altering and achieving the required collective behavior of the swarm is virtual physics where the robots are subjected to virtual forces, making them assume some spatial organization like patterns, and collective motion.

As of now, a limitation in swarm robotics is that the collective behavior of the swarm is completely dependent on the designer’s creativity, foresight and expertise. A fixed, general way to swarm building hasn’t been devised, which is why it has been proposed, that gamers’ brainwaves be used to blueprint patterns for topology construction, military applications and many more.

The automatic designs are mainly performed through the evolutionary robotics approach, where the swarm alters its collective behavior, or, evolves, to suit a certain situational condition it is subjected to.

Image depicts the ability of the swarm to create patterns from roboticsandautomationnews.com

The following table shows the advantages of this field of robotics as compared to other multi agent systems:

Applications:

Swarm robotics has a wide field of operation and can be used for making many tasks easy and automated. These techniques can be used in miniaturization, construction (for example, strengthening a part of a structure), repairs where people cannot reach or go, internal surgeries of the human body and even maybe an army of automated drone swarms. Robot swarms can either be categorized into the following two levels: The microscopic level where the behaviors of the individual robots is modeled; or the macroscopic level, where the collective behavior of the swarm is constructed.

The microscopic level of modeling allows for miniaturization of these swarm robots to nanobot levels and allows for a wide range of scalability. It allows the detailed representation of the robots, which is difficult to achieve, and most of this level of modeling is done through computer simulations.

One of the most common macroscopic modeling approaches is the use of rate or differential equations. Rate equations describe the time evolution of the ratio of robots in a particular state, that is, the evolution of activities the robots are performing in a specific area of the environment. Rate equations have been used to model many collective behaviors, including object clustering and adaptive foraging

Hybrid modeling can be done through the Fokker-Planck and Langevin equations where the individual behaviors can be modeled along with the collective functioning of the swarm.

Image depicting Nano sized swarm robots in the bloodstream from cnnhealth

Perhaps the most interesting application for miniaturized swarm technology is using it for the internal treatment of the human body. Swarms can be introduced into the bloodstream in the form of nanobots and can be used for surgeries of minute detail that need to be done internally and intricately. These robots should be very efficient and small, so as to not cause unwanted internal bleeding and thus worsen the problems. They must be properly tested and designed before implemented in the healthcare industry, but on perfection, the technology can prove to be extremely valuable to the future of healthcare and lifestyle.

Some of the applications of swarm robotics are very valued and can be referred to in the following:

For more information on internal surgeries, or Nano swarms in our blood stream, refer:

https://edition.cnn.com/2018/09/04/health/nano-swarm-robots-intl/index.html

https://www.newscientist.com/article/2141595-tiny-robots-swim-the-front-crawl-through-your-veins/

https://www.researchgate.net/publication/314612977_Navigation_and_Cooperative_Control_for_Nanorobots_in_the_Bloodstream_Environment_Based_on_Swarm_Intelligence

For more information on how Gamers’ minds can be used for swarm patterning, refer:

https://mindmatters.ai/2019/05/swarm-printing-are-ai-robots-tomorrows-construction-workers/

https://www.analyticsinsight.net/darpas-initiative-train-military-robot-swarm-using-gamers-brain/

References:

https://www.hindawi.com/journals/isrn/2013/608164/

http://www.scholarpedia.org/article/Swarm_robotics

http://www.scholarpedia.org/article/Swarm_intelligence