Optimal multi-UAV path planning in a heterogeneous 3D environment using an SVM-guided hybrid metaheuristic framework along with the B-spline technique

This research proposes a path-planning solution for a multi-unmanned aerial vehicle operating in a three-dimensional constraint environment. A novel penalty-based cost function has been developed. The proposed cost function integrates weighted sub-components, including path length, obstacle avoidance, altitude constraints, climb angle, turn angle, synchronization coefficient, and wind disturbance. A support vector machine regressor is used to find the weights of the sub-components. A hybrid optimization approach, combining the sine cosine algorithm with gray wolf optimization, is employed to optimize the proposed cost function. In this framework, the gray wolf optimization facilitates local refinement, while the sine cosine algorithm provides global exploration. The B-spline method is applied to ensure smooth trajectory generation. The results reveal that the sine cosine algorithm–gray wolf optimization–B-spline approach outperforms several other possible approaches in terms of fitness value, accuracy, and trajectory smoothness. The combined sine cosine algorithm-gray wolf optimization–B-spline approach brings together learning-based weight adaptation, structured hybrid optimization, and constraint-aware smoothing, leading to a reliable and effective solution for path planning in real-world scenarios.