Evolutionary Algorithms for Multi-Objective Optimization of Drone Controller Parameters

Drones are effective for reducing human activity and interactions by performing tasks such as exploring and inspecting new environments, monitoring resources and delivering packages. Drones need a controller to maintain stability and to reach their goal. The most well-known drone controllers are proportional-integral-derivative (PID) and proportional-derivative (PD) controllers. However, the controller parameters need to be tuned and optimized. In this paper, we introduce the use of two evolutionary algorithms, biogeography-based optimization (BBO) and particle swarm optimization (PSO), for multi-objective optimization (MOO) to tune the parameters of the PD controller of a drone. The combination of MOO, BBO, and PSO results in various methods for optimization: vector evaluated BBO and PSO, denoted as VEBBO and VEPSO; and non-dominated sorting BBO and PSO, denoted as NSBBO and NSPSO. The multi-objective cost function is based on tracking errors for the four states of the system. Two criteria for evaluating the Pareto fronts of the optimization methods, normalized hypervolume and relative coverage, are used to compare performance. Results show that NSBBO generally performs better than the other methods.