The International Journal of Robotics Research Study, Ahead of Publish.
We provide a formula that, provided a depiction of a roadway network in lane-level information, calculates a course that lessens the anticipated expense to get to a provided location. In doing so, our formula permits us to address for the facility compromises experienced when attempting to determine not simply which roadways to comply with, however likewise when to alter in between the lanes comprising these roadways, in order to– for instance– minimize the chance of missing out on a left leave while not needlessly driving in the leftmost lane. This transmitting issue can normally be developed as a Markov Choice Refine (MDP), in which lane adjustment activities have stochastic end results. Nonetheless, MDPs are recognized to be taxing to address generally. In this paper, we reveal that– under practical presumptions– we can make use of a Dijkstra-like strategy to address this stochastic issue, and take advantage of its reliable O(n log n) running time. This allows an independent automobile to show lane-selection habits as it successfully prepares an ideal course to its location.
发布者：Mitchell Jones，转转请注明出处：https://robotalks.cn/lane-level-route-planning-for-autonomous-vehicles-2/
