EU project ASSISIbf
In order to allow information to be exchange bi-directionally between a group of living honeybees (in Graz, Austria) and fish (in Lausanne, Switzerland) two species of bio-mimetic and bio-compatible robots have to be created and integrated into each of the two target societies of animals. The rest is rather easy: connect the two robots and the bees talk to the fish and vice versa.
ASSISIbf: Robots Unveil the Social Behavior Language of Bees & Fish
The key objective of the EU-FET project ASSISIbf is simple: (1) Developing two types of biomimetic robots that are capable of subtly integrating themselves into living animal societies (honeybees and fish) in a way that they take part in the swarm intelligent decision making of these animals. (2) Then connect the two robot types over the internet, even over a large distance like Graz (Austria) to Lausanne (Switzerland), so that they coordinate their actions appropriately with each other. (3) Voila, a novel biohybrid mixed-species system of bees and fish and robots emerges, in which bees and fish make their collective decisions now together, as the first mixed-species swarm, formed by mediating biomimetic robots.
This experiment was likely not only the first time that two animal species exchanged information with each other via robotic surrogates. It is also the first proof of principle that robotic mediators and surrogates might create novel ecological linkages between species, a principle I call "Ecosystem Hacking".
My research in the EU-FET project ASSISIbf is in cooperation with the following international and interdisciplinary partners: Prof. Francesco Mondada (École Polytechnique Fédérale de Lausanne, CH), Prof. Stjepan Bogdan (LARICS, Univ. Zagreb, Croatia), Prof. Luis Correia (University of Lisbon, Portugal), Dr. Serge Kernbach (Cybertronica Research, Stuttgart, D) and Prof. Jose Halloy (University Paris-Diderot, Paris-7, France). In my lab, Mag. Martina Szopek, Dr. Ronald Thenius, Mag. Michael Bodi, Sarah Schönwetter-Fuchs, Stefan Schönwetter-Fuchs-Schistek and Mag. Martin Stefanec conduct significant work in this project.
Create an array of many (up to 64) robots that can sense bees, that can affect bee' behaviors and that communicate with each other locally
This robotic array is covered with a wax sheet layer and put in a climatized and dark room so that bees feel almost like in their hive
The large array can be split up in many parallel compartments, all containing robots that can interact, building a huge "biohybrid machine" to investigate swarm intelligent collective behavior.
From the interaction of bees amongst each other but also with the robots in the interaction loop, coordinated decisions can arise in the form of spatiotemporal patterns
Bees can move closer to some robotic nodes. These nodes can sense this and make the environment even more attractive by warming it up: A positive feedback loop emerges.
Positive feedback leads to symmetry breaking in the system. Thus, the bees choose one out of two or several options. This is the info that can be passed over to fish robots then.
An arena with 4 options for the bees to choose from for a suitable aggregation spot
Each robot has a tiny computer underneath the way floor: These electronics produce heat that needs to be controlled and transported away.
Robots can detect local bee presence and density with tiny IR sensors
The ASSISIbf system can explore stimuli-patterns with machine learning algorithms (ECA: Evolutionary Computation Algorithms).
These ECA technologies can tune the local and microscopic stimuli patterns until the desired macroscopic behavior is observed.
Artistic depiction of the red-light atmosphere produced in the darkroom of the ASSISIbf arena system
Via the ASSISIbf robots real bees can also self-organize with simulated virtual bees in cyberspace
An artistic depiction of the system
An artistic depiction of the system
Hot Bees Shaking in the Spotlight
For our research with honeybees it is essential for us to investigate how local stimuli and cues affect the behavior of honeybees individually, in order to then construct models of collective behaviors of larger groups of these animals. This way, findings on mechanisms that reside on the microscopic level of the system are informing our predictions of phenomena that will appear on the macroscopic system level.
Temperature cues tested with a simple heating pad
Vibrations cues tested with a simple excenter motor
Array of robots emitting many stimuli patterns in parallel