Decentralized task allocation for multiple UAVs with task execution uncertainties

Abstract

This work builds on a robust decentralized task allocation algorithm to address the multiple unmanned aerial vehicle (UAV) surveillance problem under task duration uncertainties. Considering the existing robust task allocation algorithm is computationally intensive and also has no optimality guarantees, this paper proposes a new robust task assignment formulation that reduces the calculation of robust scores and provides a certain theoretical guarantee of optimality. In the proposed method, the Markov model is introduced to describe the impact of uncertain parameters on task rewards and the expected score function is reformulated as the utility function of the states in the Markov model. Through providing the high-precision expected marginal gain of tasks, the task assignment gains a better accumulative score than the state of arts robust algorithms do. Besides, this algorithm is proven to be convergent and could reach a prior optimality guarantee of at least 50%. Numerical Simulations demonstrate the performance improvement of the proposed method compared with basic CBBA, robust extension to CBBA and cost-benefit greedy algorithm.

Extensions of Receding Horizon Task Assignment for Area Coverage in Dynamic Environments

“…The failure rate of UAVs was considered a random variable with an approximate range rather than an exact probability distribution. Liu et al introduced robust task assignment based on the Markov decision process in the stochastic environment where the duration of task execution is uncertain due to load capacity [16]. The task duration distribution was also assumed to follow the normal distribution, while the actual distribution is unknown and can be obtained from the analysis of historical data.…”

Autonomous Search and Rescue with Modeling and Simulation and Metrics

“…More recently, research has developed abilities for decentralized tasking of multiple assets in the presense of uncertainty (Liu, Seo, Yan, & Tsourdos, 2020); however, this work did not consider core abilities to self task and generate new tasks, which will be key to future AI operations.…”