In order toperceive cooperativelya connected autonomous vehicle(CAV)

the traditional method is to fuse its obstacle list

but it can only fuse the objects that already exist in the obstacle list and cannot fuse the obstacles that are undetected by a bicycle. In order to improve the perception and cooperation ability of the CAV

this paper proposes a multi-vehicle shared geometric map generation and fusion method

which can extract the original height feature of the CAVfrom the 3D point cloud in each Pillar area during target detection with a laser radar. The height feature is compressed to obtain a 2.5D geometric map

which is combined with the results of the target detection algorithm to generate shared data and sent to other vehicles within the communication range to supplement the height information onan unknown obstacle

expand the perception range of the bicycle and reduce undetected zones. Experimental results show that the vehicletarget detection accuracy after shared data fusionreaches up to 85.71%

which is 14.28% higher than that of single-vehicle perception. Under the condition of 4G communication

the average transmission delay of shared data is 5.23 ms.