Fault Tolerant Dynamic Task Assignment for UAV-based Search Teams

TL;DR

This paper introduces a comprehensive framework for dynamic UAV task assignment that accounts for faults, battery management, and real-time decision-making using stochastic dynamic programming, enhancing reliability in search operations.

Contribution

It presents a novel integrated model combining fault tolerance, battery-aware task distribution, and stochastic optimization for UAV-based search teams.

Findings

Robustness demonstrated in fault-prone simulations

Effective battery management improves mission duration

Optimal policies can be computed offline for rapid online deployment

Abstract

This research offers a novel framework for dynamic task assignment for unmanned aerial vehicles (UAVs) in cooperative search settings. Notably, it incorporates post-fault UAV capabilities into job assignment techniques, assuring operational dependability in the event of sensor and actuator failures. A significant innovation is the utilization of UAV battery charge to assess range relative to search objectives, hence improving job distribution while conserving battery life. This model integrates repair, recharge, and stochastic goal recurrence, hence increasing its real-world applicability. Using stochastic dynamic programming, this method makes it simpler to determine optimal assignment policies offline so they may be implemented rapidly online. This paper emphasizes the holistic aspect of the proposed model, which connects high-level task rules to low-level control capabilities. A simulation-based case study proves its usefulness, highlighting its robustness in fault-prone and battery-variable settings. Overall, this paper proposes and demonstrates a comprehensive method for assigning UAV tasks that integrates defect awareness, battery management, and multilayer control through the use of stochastic dynamic programming.