A task planning method for multi-UAV from local LTL specifications

Abstract

In this paper, a receding horizon task planning method for multi-UAV systems from local linear temporal logic (LTL) specifications under probabilistic model checking is proposed, which aims to satisfy the rich task specification assigned to each UAV. Firstly, considering the uncertainties in the real scenario, the motion of each UAV is modeled as a Markov decision process (MDP). Secondly, the task specification of each UAV is expressed as a linear temporal logic formula. The tasks with the collaboration requirements of multiple UAVs take the form of atomic proposition into the LTL specifications. And the LTL specifications are transformed to deterministic Rabin automatons over which a task progression reward is defined to determine the local goal states in the finite-horizon product systems. Thirdly, two horizons are set to determine the running steps in automatons and MDPs. By dynamically building local finite-horizon product systems, the collaboration plans are synthesized iteratively for each UAV to satisfy the task specifications with maximum probability. Finally, through simulation experiments based on ROS and Gazebo, the results show that the method can synthesize correct plans for multi-UAV systems and greatly reduce the computational burden.