This paper proposes an online algorithmic framework for solving the problem of unmanned-aerial-vehicle (UAV) trajectory planning while avoiding partially unknown geofences under an uncertain environment. The real-time perception of the environment is leveraged to enhance both safety and efficiency of the generation of UAV trajectories. More specifically, to tackle with uncertainties presented in the UAV’s real-time measurements, we utilize a novel notion of dummy upper confidence bound, which was proposed in our previous work. As a result, two remarkable features are guaranteed for our online trajectory planner: 1) it is guaranteed with high probability that the regions generated by the confidence bounds are larger than the actual geofences, and thus safety of the planned trajectories is naturally assured; and 2) the confidence regions shrink progressively as more perception information is acquired, which further improves the efficiency of the UAV’s trajectories. With these two features, both feasibility and (sub)optimality are theoretically ensured for our online trajectory planning algorithm. At last, two sets of numerical examples are provided to demonstrate all theoretical findings in the paper.
Unmanned-Aerial-Vehicle Online Trajectory Planning Using Confidence Bounds of Chance-Constrained Geofences
Journal of Guidance, Control, and Dynamics ; 48 , 1 ; 115-126
01.01.2025
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch