近幾年,機器人相關領域漸漸受到矚目。SLAM這個領域已經被研究一段時間,是一項機器人完成高難度任務所需的基礎。然而,感測器融合的重要性在近幾年才開始被研究。 光達與慣性感測器融合之定位與建圖系統在這篇論文中被提出,其中也包含回環偵測及全局位姿優化。光達與慣性感測器被使用來互相補足對方的缺點,達到更好的位置與姿態估測。在SLAM系統中,估測出的位姿及地圖提供機器人環境資訊,使其實現搜索式路徑規劃。在同時考慮環境中障礙物與自身的運動限制下,搜索式路徑規劃可以產生出一條平滑且最短時間的軌跡。 最後,這個光達與慣性感測器系統在KITTI資料集中測是,以評估其方法之精確度。同時也在室內環境中,在無人機上實現搜索式路徑規劃,達到實時地產生一條局部最佳的路徑。
In recent years, the field of robotics has attracted lots of attention. For the robot to perform high level tasks, SLAM becomes a fundamental technique to build on. The SLAM problem has been well studied for a period of time, but the importance of sensor fusion to complement each other was just investigated in the past ten years. In this work, a tightly-coupled lidar-inertial SLAM system is developed with loop-closure and pose graph optimization. The estimate pose and map from the SLAM system provide the knowledge of the environment for the search-based motion planning method. By taking the obstacles in the surrounding and the motion constraints into consideration, a smooth and minimum-time trajectory is generated. Finally, the lidar-inertial SLAM system is evaluated in KITTI dataset, and an indoor flight experiment has proven the capability of generating a locally-optimal trajectory in real-time.