Research On Field Ground Segmentation Method Based On ToF And Application In Simulation Platform

The field combat environment is complex and changeable,and in the research on the environment perception with rugged terrain and unknown obstacles of unmanned combat platforms,how to improve the precision of ground area segmentation where vehicles can travel is the first problem to be solved.Previous studies on ground segmentation problem is usually with the aid of visible light camera(typical passive sensing)or LiDAR(typical active sensing),but the former is sensitive to environmental conditions such as light and texture,the latter will cause scanning deformation easily in a bumpy environment because of its scanning mechanism,thus they are not stable to adapt to the field environment.The appearance of imaging sensor based on time-offlight(ToF)range technique greatly improves the machine's ability to perceive the environment.ToF camera can obtain the depth information of vehicle's front field environment through one imaging,and compared with the binocular/multi-ocular stereo vision system based on visible light camera and LiDAR,it can better adapt to the field environment.Therefore,this paper will conduct research on field ground segmentation based on ToF ranging technology.Aiming at the requirements of unmanned combat platform for environment perception,this paper proposes and designs a ground segmentation method based on ToF technology to obtain three-dimensional information of environment.Then,for the research and development of ToF camera and the ToF application research in the field of ground segmentation,the ToF sensing application simulation platform is designed and implemented,and the ground segmentation algorithm is simulated and verified on the platform.The main work and research results of this paper include the following three aspects:1.The imaging sensor based on ToF range technology was applied to field ground segmentation for the first time,and the ground segmentation method based on sizeadaptive grid was designed and implemented,that adjusted grid resolution adaptively according to the traversability analysis of terrain.Experiments show that this method can effectively segment the ground in multi-obstacle,few-obstacle and barrier-free environments,and can effectively improve the precision and recall rate while comprehensively considering the precision and execution efficiency of the segmentation algorithm.2.Aiming at the need of ToF perception research in rugged off-road terrain,an ToF sensing application simulation platform is designed and implemented.The geometric model of ToF camera imaging is analyzed,and the diamond-square fractal algorithm is introduced to randomly generate natural rugged terrain.The simulation of ToF camera imaging under the simulated terrain environment is realized,and the ToF 3D imaging data under ideal conditions can be obtained to support the perception research experiment.3.Based on the secondary development interface designed on the simulation platform,the field ground segmentation method based on the size-adaptive grid proposed in this paper is realized.Through application experiments,the effectiveness of the ground segmentation method in this paper is proved,and the influence of main parameters of ToF camera such as the effective range on the ground segmentation effect is analyzed.