An improved butterfly optimization algorithm for UAV path planning in complex environment

The three-dimensional (3-D) path planning of unmanned aerial vehicles (UAV) is a multi-objective optimization problem. In this work, an improved butterfly optimization algorithm (IBOA) based on the virtual center butterfly (VCB) and Neighborhood dimension perturbation learning (NDPL) is proposed to solve the problem. Since BOA executes a global search using pairwise interactions between two random butterflies (random operators). It makes the algorithm susceptible to missing the optimal solution, resulting in a lack of exploration capability. Therefore, we designed a novel VCB strategy to improve the exploration capability of the algorithm by creating attraction or repulsion effects on butterflies during the global phase. In the exploitation phase, the other individuals gravitate toward the best individual. If the best individual falls into the local extremes, the algorithm converges prematurely and with low precision. A new NDPL method is proposed that enhances the local mining capability of the algorithm by constructing a neighborhood matrix and learning from individuals in the neighborhood. The results of the simulation experiment in four scenarios demonstrate that the IBOA can effectively acquire a practical and effective path, and it performs better than the other six algorithms.