Back to the program.
Robots, Computer Vision and Machine Learning: New Tools for the Analysis of Honeybee Communication
Prof. Tim Landgraf
Biorobotics Lab
tim.landgraf@fu-berlin.de

Abstract

Honeybees are a popular model in a variety of research fields, such as navigation, communication and collective intelligence. Bees are versatile learners and exhibit remarkable cognitive capabilites on the individual but also show remarkable adaptive performance on the colony level. A well known example of extraordinary individual performance is the honeybee dance in which a forager bee signals the location of a valueable food source to her nestmates through stereotypical body movements. Dance followers decode the information contained in these movements and use them to find the advertised field locations even several kilometres away. To put our understanding of the dance to the ultimate test, we have built a honeybee robot capable of imitating the dance in its various components. In field experiments we were able to attract bees towards the robotic dance, excite the dance-following behavior and recruit bees to remote field sites. For some animals we were also able to track their consecutive flight via harmon ic radar. Even though "RoboBee" is able to recruit bees, its recruitment efficiency is low compared to natural dances. In order to investigate how the dance is embedded in the colony's various other communication behaviors, and how individual experience shapes dance-related preferences, we have built a long-term tracking system for all bees in a colony. I will review the current prototype, first experimental results and give an outlook towards how this system will help in understanding how information in general is integrated into the colony by each individual worker, how information flows through the social network, how the colony processes information and eventually computes a meaningful output.

Keywords

swarm intelligence, swarm robotics