Fault tolerant mission management for UAV under random threat using Markov decision process

Abstract

This research introduces a comprehensive framework for mission accomplishment of unmanned aerial vehicles (UAVs) operating in threat-prone zones. Leveraging a Markov Decision Process (MDP), the proposed model ensures mission accomplishment and optimal resource utilization by incorporating UAV state-of-charge, post-fault capabilities, and threat navigation strategies. The framework addresses sensor, actuator, and vision camera faults, enabling dependable operations even under adverse conditions. A key feature of the model is the integration of UAV battery levels to evaluate operational range relative to mission objectives, optimizing task distribution while conserving energy. Additionally, the incorporation of adaptive navigation modes enhances UAV agility and robustness by enabling threat avoidance during operation, closely emulating real-world scenarios. By synthesizing repair protocols, recharging strategies, and stochastic modeling of recurrent goals and threats, the framework offers a holistic solution to improve resilience and mission success in hostile environments. Stochastic dynamic programming ensures the rapid application of precomputed optimal policies during mission execution. A simulation-based case study demonstrates the framework's effectiveness in navigating threats, mitigating faults, and ensuring mission reliability in energy-constrained scenarios.