Resilient search-attack task planning for heterogeneous unmanned systems in adversarial environments

In this study, a resilient task planning method is introduced to enable heterogeneous unmanned systems (USs, e.g., unmanned aerial vehicles) to search and attack targets in dynamic and adversarial environments. To overcome the operational performance degradation caused by hostile attacks or trajectory deviations, a resilience-based task allocation (RTA) method is proposed, which can dynamically match the heterogeneous USs with specific requirements of searching designated regions and attacking detected targets by monitoring the task performance of the USs using a designed evaluation metric. To reactively deal with pop-up events, such as the appearance of new targets or threats, the event-based task execution (ETE) method is developed that explicitly defines the event-based signal temporal logic (STL) and the STL-based control barrier functions for the control synthesis of the USs. This designed control strategy can directly transfer complex task specifications to executable control inputs, which is scalable to accommodate more events by adding constraints to a quadratic programming problem. By setting multiple moving targets with denial ability scattered in a given area, simulated experiments demonstrate the efficacy and resilience of the proposed method, outperforming two conventional methods with a higher rate of target detection and destruction.