Distributed planning of multi-agent systems with coupled temporal logic specifications

In this paper, we investigate the multi-agent mission planning and control problem under coordination constraints such as collision avoidance and spatial coherence. We propose a sequential trajectory planning algorithm that is run by each agent in a distributed manner. Mission requirements including servicing specific regions, avoiding static and dynamic obstacles (other agents), and flying close to the other agents are incorporated into the problem as Signal Temporal Logic (STL) constraints. The planning problem under STL constraints is formulated as a Mixed Integer Linear Program and an off-the-shelf tool is used to find the high-level trajectories. The resulting high-level trajectories are then tracked by the agents via Model Predictive Control. Illustrative simulation and experiment results are presented in addition to comparing the proposed method with a centralized solution.