This project bridges the gap between our honeybee aggregation experiments and our general swarm robotic experiments. By adapting our Hemisson robots to work with temperature sensors we want to emulate the honeybee behaviour even more realistically. These robots use two antennae-like temperature sensors to find the place with the optimal temperature inside a small arena. To do this they use our BEECLUST algorithm, which was inspired by the clustering behaviour of young honeybees inside a temperature arena. Robots that use this BEECLUST algorithm can be described as being "swarm-intelligent" because the algorithm only works with multiple robots and increases its efficiency with increasing number of robots. The advantages of this approach are that this form of collective optimum finding is simple and nevertheless quite robust and flexible.