Discovery of parameters for animation of midge swarms

Abstract

We describe a method of discovering suitable parameters for simulating and animating the swarming behaviour of non-biting midges (Diptera: Chironomidae). A characteristic of animal aggregations that can be emulated by software is the emergence of complex behaviours from simple rules. Here the well-characterized swarming behaviour of non-biting midges is used to create a rule-based behaviour model for them. To test the effectiveness of this model in creating the emergent qualities of real swarms, success criteria are derived from quantitative swarm data. We propose using a genetic algorithm to automate the identification of parameter settings that optimize the effectiveness of the model.