Robust path planning for autonomous vehicle in position uncertainty

The abrupt and frequent change of a path for autonomous vehicle might occur in the presence of position uncertainty due to sensor noises or errors in localization. While path planning of an autonomous vehicle utilize multiple sensors to generate the path, it is necessary to not only reduce such errors but also to plan a consistent path. Therefore, we propose a framework that can generate a stable path by considering the history of previous paths. The result of experimental simulations indicates the decrease in rate of switching path directions in three different scenarios in position uncertainty. This framework demonstrates the effectiveness of path generation by applying a history parameter to modify cost function of the A* algorithm.