Take the collection of typical road unevenness in the orchard as the research object，establish a road roughness acquisition method based on lidar point cloud processing，built a road roughness acquisition system platform based on 3D lidar，combine point cloud processing technology to complete the extraction of road elevation information.By using AR （autoregressive） model to calculate the road power spectral density based on the proportional analysis method to determine the roughness level，and through the acceleration vibration recorder for system verification，using the system to carry out a typical orchard road unevenness data collection test. The test results show that the roughness of the orchard road surface showed that the cement pavement is mainly concentrated in the B grade，and the B grade accounts for 82.33%. The sand and gravel pavement is mainly concentrated in the C grade，and the C grade accounts for 84.00%. The mud pavement is mainly concentrated in the D，E grade，with D grade accounting for 48.67% and E grade accounting for 31.00%. The results of roughness evaluation indicated that the application of the three-dimensional lidar orchard road roughness acquisition system is reliable and the results of evaluation are accurate. It is suitable for collecting level of road roughness in mountain forests，fruit，and tea gardens.