Development, evolution and genetic analysis of C. elegans-inspired foraging algorithms under different environmental conditions

In this work 3 minimalist bio-inspired foraging algorithms based on C. elegans’ chemotaxis and foraging behaviour were developed and investigated. The main goal of the work is to apply the algorithms to robots with limited sensing capabilities. The refined versions of these algorithms were developed and optimised in 22 different environments. The results were processed using a novel set of techniques presented here, named Genotype Clustering. The results lead to two distinct conclusions, one practical and one more academic. From a practical perspective, the results suggest that, when suitably tuned, minimalist C. elegans-inspired foraging algorithms can lead to effective navigation to unknown targets even in the presence of repellents and under the influence of a significant sensor noise. From an academic perspective, the work demonstrates that even simple models can serve as an interesting and informative testbed for exploring fundamental evolutionary principles. The simulated robots were grounded in real hardware parameters, aiming at future application of the foraging algorithms in real robots. Another achievement of the project was the development of the simulation framework that provides a simple yet flexible program for the development and optimisation of behavioural algorithms.