论文部分内容阅读
目的:针对仅能通过轻型激光测距仪获取半稠密点云的环境地图构建问题,提出一种构建稠密表面模型的方法。该方法使机器人能够利用所构建的稠密表面模型地图完成定位、导航及目标搜索等任务。创新点:提出一种基于点云分割的点云表面重采样方法及一种基于点云概率模型的表面模型融合方法。对半稠密点云进行保留表面结构特性的重采样来获取观测数据的稠密表面模型。并递增式地将新获得的稠密表面模型融合进已有的稠密表面地图中,从而获得几何一致性较好的环境表面模型地图。实验效果:图6、7展示了基于本文方法所构建的稠密表面模型地图的效果。其几何结构精确且表面纹理清晰。此外,图8、9分别重点展示了表面重采样的作用以及本文提出的重采样方法的效果。图11则展示了本文方法对表面模型动态更新的较好支持。结论:使用本文所提方法,机器人可携带轻便式激光测距仪,获取半稠密点云后再进一步处理和融合得到几何一致性较高、表面精细的稠密表面问题模型地图,更好地实现定位、导航及目标搜索等任务。
OBJECTIVE: To solve the problem of constructing an environmental map that can only obtain semi-dense point clouds by light laser range finder, a method of constructing dense surface model is proposed. The method enables the robot to utilize the constructed dense surface model map to accomplish such tasks as positioning, navigation and target search. Innovative point: A point cloud surface resampling method based on point cloud segmentation and a surface model fusion method based on point cloud probability model are proposed. A dense surface model was obtained for the semi-dense point cloud by preserving the resurfacing of the surface structural properties to obtain the observed data. And the newly obtained dense surface model is incrementally incorporated into the existing dense surface map to obtain the geometrically consistent environment surface model map. Experimental results: Figures 6 and 7 show the effect of a dense surface model map constructed based on our method. Its geometric structure is accurate and the surface texture is clear. In addition, Figs. 8 and 9 respectively highlight the effect of surface resampling and the resampling method proposed in this paper. Figure 11 shows the better support of this method for dynamic updating of surface models. Conclusion: Using the method proposed in this paper, the robot can carry a portable laser range finder to obtain semi-dense point clouds and then further processed and fused to obtain a model map of dense geometric surface problems with high geometric consistency and fine surface to better locate the position , Navigation and target search tasks.