Bayesian optimization to enhance coverage performance of a swarm of UAV with chaotic dynamics

We introduce the optimization of CACOC through Bayesian Optimization. CACOC is based on a chaotic system, i.e. Rossler system whose behavior can be modified by tuning the α parameter. In order to evaluate the performance of CACOC for different value of α, the coverage metric has to be evaluated after simulation. The latter is time-consuming. Therefore, a surrogate-based optimization, i.e. Bayesian Optimization has been privilegied to tackle this issue. An analysis of the chaotic system with the obtained α value has been performed to compare the periodic orbits and their associated patterns. Numerical results show that the best α value avoid a waste of time in periodic region of the bifurcation diagram. Future works will focus on more complex chaotic system as well as new application domain of the optimized CACOC approach.