Home > Research Groups > Algorithms and Complexity > Research > Swarm Intelligence and Evolutionary Robotics

Swarm Intelligence and Evolutionary Robotics

The design of adaptive systems which are able to operate autonomously and reliably in dynamic environments, is one of the big challenges in robotics. The methods of swarm robotics help to reduce the complexity of such systems and methods of evolutionary robotics help to generate robot controllers automatically. In our project “flora robotica”, we apply these methods to investigate novel approaches in combining robots and natural plants.

Swarm Intelligence

Swarm Intelligence is the discipline that deals with natural and artificial systems composed of many individuals that coordinate using decentralised control and self-organization. In particular, the discipline focuses on the collective behaviours that result from the local interactions of the individuals with each other and with their environment. Examples of systems studied by swarm intelligence are multi-robot systems and certain computer programs that are written to tackle optimisation and data analysis problems. The objective of our approach is to define individual behaviours based on simple algorithms but still generate complex system behaviours. In this way we hope to be able to govern the increasing complexity of engineered system also in the future.

Evolutionary Algorithms

Evolutionary Algorithms belong to the field of meta-heuristic optimisation and they are based on the Darwinian principle of selection and reproduction of the well adapted organism. We apply this method in software engineering for the automatic synthesis of requirements specifications and in robotics to generate controllers for autonomous robots automatically.

The latter approach considers robots as autonomous, artificial organisms that develop their own skills without human action and closely interacting with their environment. The objective is to get rid of both the manual work of describing requirements specifications and the manual programming work of robot controllers.

Vision of the “flora robotica” system

flora robotica

This year, we have accomplished an essential step of development in the EU-funded project “flora robotica”. Since April 2016, we have been in the second project year. The idea is to create closely linked symbiotic relationships between a distributed robot system and natural plants to produce architectural artefacts and living spaces.

We have developed and tested the first prototypes of the required robot hardware. In different experiments we have investigated how plant growth can be changed arbitrarily by the influence of light, vibrations, and mechanical stimuli. To do so, we activate the plant’s natural tropisms (growth and motion of the plant as reactions to different stimuli) as desired. Selecting appropriate sensors to detect plants, adjusting ambient light and directed light while respecting the plant’s needs, and the software-hardware integration have been special challenges.

Overall, the project is a prime example of the role of modern computer science in interplay with the life sciences, even architecture, and ideas from the Internet of Things as well as Ubiquitous Computing.

Robot hardware for the control of plants and concept of braided structures.

Supported by:

  • DFG Collaborative Research Centre 901 “On-The-Fly Computing”, subproject B1; EU Horizon 2020 FET project flora robotica