Hybrid particle swarm unbiased extremum seeking control for cooperative drone fleet for gas leak localization in obstructed outdoor environments

The study employs a fleet of drones to address gas source localization (GSL) in a dynamic outdoor environment. A hybrid control framework integrating particle swarm optimization (PSO) with unbiased extremum seeking (uES) is proposed to improve GSL performance under turbulent plume conditions. The proposed framework combines the global exploration capability of PSO with the smooth, unbiased gradient refinement of uES to address the limitations of existing algorithms. Comprehensive assessments conducted in two simulated plume environments, along with high-fidelity Gazebo– GADEN simulations, indicate that PSO–uES reliably achieves fast convergence, high success rates, and notably stable trajectories. Unlike classical ESC, which exhibits star-shaped dithering and does not reliably settle near the source, and standard PSO, which may wander within areas of weak gradients, the proposed approach sustains directed movement, minimizes plume-loss occurrences, and converges close to the source location in simulation environments. These findings demonstrate that PSO–uES is a promising approach for simulation-based GSL with complex and turbulent plume characteristics.