How to achieve Swarm Intelligence

Eric Monchalin

Vice-President, Head of Machine Intelligence at Atos and member of the Scientific Community

Purshottam Purswani

Chief Architect, Distinguished Expert and member of the Scientific Community

Do van Rijn

Bid Executive at Atos

Posted on: 27 February 2020

In the previous blog article in our series on Swarm Intelligence, we illustrated why it’s high time for organizations to create man-made swarm intelligence based not on what we can extract from nature, but on what we can learn from her.

Let’s discuss in more details how to develop a swarm intelligent system.

Tips to achieve Swarm Intelligence

Few key principles to keep in mind when designing swarm intelligence agents are:

▶    Goal seeking
All agents are designed to work toward the same objective

▶    Control loops
Stay focused to achieve the task it has been assigned

▶    Openness
Every agent responds to the conditions and the events encountered

▶    Memory and knowledge sharing
Rely on communication between adjacent individuals that maintains local knowledge shared between neighbors

▶    Parallelism
Define tasks which can be decomposed and done in parallel

▶    Fault tolerance
Design system where a failure of a single agent does not prevent a given task to be accomplished without significantly degrading performance

▶    Scalable
Allow agents to dynamically join or quit a task at any time without interrupting the whole swarm.

Swarm Intelligence in action

Based on these design principles, we already see applications designed and developed. Let’s look at the example of the Wind turbine farm. Here the swarm capabilities apply in various areas, leveraging the capabilities of the algorithmic school:

·         Smoothing of wind power fluctuations;

·         Optimization of the rotor speed and tip-speed ratio to maximize power and energy capture from the wind;

·         Smoothing of wind power fluctuations.

It also perfectly matches a digital twin approach as we discussed in our blog  article “Swarm Intelligence, a driving force in our ultra-connected world”.

Lastly, Swarm Intelligence is key for tomorrow’s smart grid mode of operation to efficiently manage local production peaks.

Swarm Intelligence is also in action in underwater habitats, where swarm of autonomous underwater vehicles are able to interact with each other and can balance tasks such as ecological monitoring, searching, maintaining, exploring and harvesting resources…

Another example is the space exploration where multiple-spacecraft missions offer greater likelihood of survival and flexibility than single ones. These scenarios have been explored by the NASA.

Pollutant absorbing is also at stake, researchers of the Massachusetts Institute of Technology have developed a fleet of low-cost oil absorbing robots called ‘Seaswarm’ for ocean-skimming and oil removal. It delivers an autonomous and low-cost solution for ocean environment protection.

Lastly, at the last Olympic Games, a swarm of drones revealed amazing distributed intelligence to build the Olympic rings.

Let’s now imagine the future possibilities of swarm applications. When autonomous cars will become a reality, Swarm Intelligence can be applied to let cars mutually determine the overall optimal traffic pattern, thus minimizing delay due to traffic jams and optimizing the average speed of all cars in a certain area. Each of these cars will still have its own specific destination to reach.

Over the last decades, a large variety of algorithms has been developed to solve such challenges. For instance, PSO (Particle Swarm Optimization), we already discussed in our blog article here is applied to solve optimization problems, i.e. find out the most appropriate solution in a set of candidate solutions. It is widely applied to resolve a large diversity of simulation, analytical and design problems.

 

In conclusion, Swarm intelligent agents need to have distributed decision making and knowledge sharing capabilities, which:

1)      Enable a decomposition and delegation of the overall problem resolution task to the agents

2)      Result in self-organization among the agents.

In turn, this leads to a collective behavior striving to achieve the same objective to solve the initial issue.

Is Swarm Intelligence going to be a game changer? Stay tuned with our next blog.

Read the full blog series here:

  1. The rise of the Swarm Intelligence era
  2. Swarm intelligence as an innovation booster
  3. Swarm Intelligence promises to make our life and business easier
  4. Swarm intelligence, a driving force in our ultra-connected world
  5. Communities swarming in the virtual world
  6. It's time to engage with nature through Swarm Intelligence

 

Anxious and excited to hear more? Read our white paper to get all you wish to know and even more.

Download the Swarm intelligence – White Paper

Share this blog article


About Eric Monchalin

Vice-President, Head of Machine Intelligence at Atos and member of the Scientific Community
Eric Monchalin is Vice-President, Head of Machine Intelligence at Atos. He’s responsible for solidifying new technological and business directions for the Big Data and Security Global Business Line. Eric’s career has been mainly built on numerous R&D positions in several companies, with experience in leading 100+ people organizations and managing multi tens millions Euros projects in international environments. He is a technology-minded person who values wide range of skills and technological knowledge focused on customer wishes to turn them into reality.

About Purshottam Purswani

Chief Architect, Distinguished Expert and member of the Scientific Community
Purshottam is a chief architect with more than 20 years cross-industry IT experience in the Telecom, Manufacturing and financial sectors. In his 20 year experience, he has fulfilled a variety of roles including Enterprise architecture, IT management, and program management. Purshottam is now responsible for the strategy and driving innovation for customers. In his role, he displays thought leadership with regard to using business technology to address the current and future challenges faced by organizations. A member of the Atos Scientific Community and part of Atos Distinguished expert, Purshottam is very passionate and works on digital technologies around IoT, Cloud, Machine Learning, Blockchain and AI. He is married with 1 child, based out of Mumbai, India.

Follow or contact Purshottam


About Do van Rijn

Bid Executive at Atos
As a Bid Executive at the Strategic Sales Engagements team of Atos in BTN, Do van Rijn is responsible for managing bid teams with the aim to acquire new profitable business. Together with the bid team, the challenge is how to differentiate Atos positively from the rest of the pack. This requires, apart from understanding the drivers and issues of the client deeply, thinking out of the box and seeing how new paradigms, like Swarm Intelligence, can be utilized for this purpose. In other words, it is about looking for opportunities which are truly transformational to the client.