In practice dynamic replanning algorithms are always launched from the goal to the robot configuration. These are the two reasons. First, it is obvious that dynamic trajectory planning algorithms are more efficient when changes appear close the goal location (Koenig et al., 2004). Since an AUV can only detect the changes that are close to its location (because of the reflection of the sonar beam by the obstacles and because of the limited range of the sonar), it is logic to consider the robot configuration as the goal configuration for the replanning algorithm.
Second, and more importantly, since the robot is continuously moving, its location is necessarily different between two consecutive replanning processes. If the robot configuration was considered as the start configuration of the replanning algorithm, the whole distance function should be permanently recomputed and the computational efficiency of the dynamic planning algorithm would be lost.
Was this article helpful?