Swarm robots

Research question

Swarms of robots have previously been shown to be capable of self-organizing into complex predetermined patterns, however, in natural systems nests are built according to functionality rather than shape. Likewise, in this workshop we will generate emergent structures that exhibit certain predictable and potentially useful features or functions instead of specific shapes and geometries given in a top-down manner. We will explore how such a construction process can be directed through embodied coding of parts and robots rather than software control. 

Project goals

This work seeks to explore the relationship between a swarm of simple robots, how they manipulate their environment and self-organise through stigmergy1. We will explore the specific features that can be achieved reliably by the system, and how the parameters of embodied coding affect them. We will use both a physical swarm platform and a custom digital simulation to explore the design space and tune the assembly process. 

Investigations

The work investigates the outcomes of the experiments either from the perspective of s​tructures.​ Our interest lies in the qualities and properties of the structures and spaces that emerge. The ambition is to understand how these properties can be controlled by altering either the shape of robots themselves, the building blocks, or the environment.

Below are examples of such properties :

STRUCTURES:Structures are evaluated through the aggregation of clusters, formed by either blocks or robots. Their morphology in terms of linearity (continuity) is also investigated.

CLUSTERING: How can we control the sizes of structures/clusters that emerge? Can we form clusters of certain qualities, like elongated, round, etc? DURABILITY: Are the spaces temporary or permanent in nature?

DURABILITY: Are the structures temporary or permanent in nature?

SPACES: Spaces are defined as voids, in-between blocks and clusters of blocks, that are travelled by the robots.

CLUSTERING: Are separated spaces formed, or is it an integrated network? To what degree are the spaces one or two dimensional? And if they are two dimensional are they large or small?

DURABILITY: Are the spaces temporary or permanent in nature?

The investigations are done to build understanding of how to alter the set-up in order to control the behaviour in any one of the above areas. For each investigation  we have attempted to come up with a clear and consistent way of measuring the performance. We did that by adjusting the simulation based on our design goal and measuring how close to this goal we could arrive when running the physical experiment. We did this by developing a simulation tool in Processing and Analysing the video captures we collected through a blob detection algorithm within Python. The variables we focused are the following: a) Shape of robots b) Number of robots (% of arena size) c) Shape of blocks  d) Type of blocks e) Number of blocks (% of arena size) Friction of blocks f) Environment, (Static elements )

Supervision:

Kirstin Petersen, Petra Jenning, Nills Napp

Team:

Evangelos Pantazis, Manos Vermisso, Jasmin Sadegh