DV-LOAM: Direct Visual LiDAR Odometry and Mapping

Self-driving cars have experienced rapid development in the past few years, and Simultaneous Localization Mapping (SLAM) is considered to be their basic capabilities. In this article, we propose a direct vision LiDAR fusion SLAM framework that consists of three modules. Firstly, two-staged visual odometry module, which frame-to-frame tracking step, an improved sliding window based thinning proposed estimate accurate pose camera while maintaining efficiency. Secondly, every time keyframe generated, dynamic objects mapping module utilized refine obtain higher positioning accuracy better robustness. Finally, Parallel Global Local Search Loop Closure Detection (PGLS-LCD) combines Bag Words (BoW) LiDAR-Iris feature applied for place recognition correct accumulated drift maintain globally consistent map. We conducted large number experiments on public dataset our mobile robot verify effectiveness each framework. Experimental results show algorithm achieves more estimation than state-of-the-art methods.