In this paper we report the first results of evolving bio-hybrid societies. Our goal is to have robots that are integrated in an animal society, and here we evolve robot controllers using animals as fitness providers, directly judging the success of integration. In particular, we are using juvenile honeybees and robots that are able to produce vibration patterns. Previous studies have shown that honeybees react to different vibration patterns, such as exhibiting freezing or stopping behaviours. In this paper we investigate whether we are able to evolve a vibration pattern that acts as a locally acting ‘stop signal’ for bees. Honeybees were placed in two containers with no communication between them: one with an active, vibrating robot, and a second with a passive robot. Post-hoc evaluations of key evolved digital genotypes generally confirm fitness values obtained during evolution. We also tested the transferability of key genotypes to a single container, in which bees are free to visit one vibrating and two dummy robots. Encouragingly, most genotypes are able to selectively stop bees, i.e., only in the vicinity of the vibrating robot, despite having been evolved under the more constrained setup. These results speak to the value of an evolutionary approach for discovering how to interact with animals.

In this paper we report our ongoing work with evolving biohybrid societies. We develop robots that will be integrated in an animal society and will be accepted as a conspecific. Moreover, we want our robots to affect the behaviour of animals. We are using evolutionary algorithms to optimise robot controllers, where fitness is evaluated via measuring the effect a robot controller has on the animals. Several issues have to be considered: if the animals do not have a homogeneous behaviour several evaluations are needed to rule out outliers, and yet evaluating animal behaviour is a time consuming task. Besides the time it takes to record their behaviour, we have to take into account animal resting time, stimulus habituation, and feeding periods. Another factor that increases the task difficulty is robot heterogeneity, which is similar to the so called reality gap problem that occurs in evolving robot controllers in simulation. In our case, if we want a robust robot controller, we have to evaluate it in different robots. Overall, we found that doing online on-board evolutionary computation with robotic devices and animals is extremely challenging and we provide clues to avoid its major pitfalls.