[POST: 01] Reading swarm robotics related literature

I got a chance to go through the book “Swarm Robotics: A Formal Approach” by Heiko Hamann (https://doi.org/10.1007/978-3-319-74528-2). Since I’m in dire need of clearly understanding what and how this swarm robot thing works, I thought of going through the book from start to end as quickly as I can. Here are some of the info I picked up from the book.

1.  How many agents should be there in a swarm

Turns out, having a threshold is not going to work as it becomes a philosophical question. For example, if 100 is set to be the lower limit to qualify as a swarm, what about a group of 99 robots? Is it fair to not call them a swarm?

“The way out of this dilemma is to prevent setting thresholds and to stay with a method-relative definition. A swarm is not necessarily defined by its size but rather by its behavior.”

In other words, how they react to the environment, how they react with each other, how they communicate, etc. needs to be taken into consideration to identify a swarm.

 

2   Characteristics of a Swarm

Communication of a swarm is going to be limited to a local space, which means an agent will communicate only with the agents around it but within a specific range. A single agent will not bother about what agents on the far side of the swarm are doing.

The control of a swarm should be decentralized (this doesn’t mean the agents will mind their own business since they have their own control). This eliminates the single point of failure. The decisions will be taken individually but after considering the status of the surrounding and communicating with the neighboring agents.

There will be a lack of synchronicity in the swarm. This doesn’t mean the swarm is not coordinated, but there will be no global clock that is shared by everyone to start or stop a specific task.

The robots are simple. This is because no individual agent is supposed to do a task alone. There will be multiple agents which add up their capabilities to increase the overall strength of the local group in the swarm. Hence the development of the agents can be done in a simple approach. (What that means is the number of sensors on the robot can be reduced, actuators with low capacity can be used, and low processing power can be given to individual robots)

 

3.     Advantages of a swarm

The following main advantages were identified from the book,

• Robustness

The increased fault tolerance and fail-safety is present in a swarm due to redundancy. Because of this nature, in most scenarios, the agents can be manufactured even with unreliable, inexpensive parts. But in some applications, this might not be a good approach.

• Flexible

Since all the robots are the same, any robot can do any task assigned to anyone. Robots can overcome limits via cooperation.

 [Source: https://simpsons.fandom.com/wiki/Sherri_and_Terri]

• Scalable

Since each robot interacts only with agents in its local neighborhood, increasing the swarm size is possible without the need to redefine the communication methods.

 

 

4.     What is not a robot swarm

The following segment gave me a clear idea of what not to do when I’m developing the algorithm,

“Multi-agent systems, however, often rely on the distribution of or access to global information, broadcasts, sophisticated communication protocols (e.g., negotiations) that require reliable robot–robot communication, explicit assignments of roles that require robots to identify individual robots or to know the total swarm size. These systems typically do not scale also because of applied, non-scalable technologies such as Bluetooth or WLAN (wireless local area network, aka WiFi)”

I will publish a new post as soon as I read enough new information to be included in the post.