Dynamic Target Pursuit by Multi-UAV Under Communication Coverage: ACO-MATD3 Approach

This study proposes a new approach for cooperative pursuit of dynamic targets under communication coverage involving multi-unmanned aerial vehicles (UAVs). This approach combines the ant colony optimization algorithm with the multiagent twin delay deep deterministic policy gradient, called ACOMATD3. The ACO-MATD3 algorithm dynamically adjusts hyperparameters based on varying stages and requirements, greatly enhancing the stability and performance of cooperative multiUAV pursuit tasks, especially under strong communication coverage. Experimental results demonstrate that the ACO-MATD3 algorithm significantly outperforms other algorithms in terms of mean reward and communication return.