Probabilistic Trajectory Planning for UAVs in Dynamic Environments
Uğur Zengin, Atilla Dogan · 2004
In this paper, we will introduce a probabilistic solution to the problem of trajectory planning for a UAV ying in a dynamic environment. By dynamic environment we mean that the probabilistic map is time-variant i.e the probability of becoming disabled at a given location might be changing over time. Probabilistic map is dened as the risk exposure to the sources of threat as a function of time and position. This might be the case when the likelihood of threat sources and/or obstacles changing their position is a priori known. The objective in the trajectory planning is to arrive at a given target position while maximizing the safety of the UAV in a feasible trajectory. By feasible we mean that the turning rate constraints and velocity constraints of the UAV are not violated along the trajectory. A locally minimizing strategy will be used in single as well as multiple-target implementations. The strategy uses the local information of the probabilistic map and the information about the location of the target. It is parameterized to change the weighting on nding a shorter path or nding a path with smaller probability of getting disabled. Since the probabilistic map is changing with time, the probability of getting disabled at a given location will be constantly changing. Thus, the paths generated by the strategy will be functions of time as well as position. This will tell the UAV not only what path to follow but also how to adjust its speed on the path while satisfying the given velocity constraints.