nothing is saved after process

[LvJohny]

Dear Liu. Thanks for your nice work. I meet some problems. And there is only a empty path.csv in folder pcd after processing. Data is collected by uav. And I have set accumulate_cloud_num to 360. Running result is below.

[ map_builder.cc: 716 ] [INFO] Finishing Remaining Computations...
[ map_builder.cc: 396 ] [INFO] point cloud thread exit.
[ map_builder.cc: 665 ] [INFO] no enough frames for new submap, quit
[ map_builder.cc: 712 ] [INFO] submap processing done.
[ map_builder.cc: 566 ] [INFO] All submaps have been connected.
[ map_builder.cc: 616 ] [INFO] generating path files ...
[ memory_manager.cc: 62 ] [INFO] End managing memory.

more

./mapping_lidar_only.sh [ static_mapping_node.cpp: 112 ] [INFO] Get point cloud from ROS topic: velodyne_points [ static_mapping_node.cpp: 126 ] [WARNING] No imu topic, expect no imu messages. [ static_mapping_node.cpp: 146 ] [WARNING] No odom data accessed! [ static_mapping_node.cpp: 161 ] [WARNING] No GPS data accessed! [ map_builder.cc: 152 ] [INFO] Got tf : tracking -> lidar translation : (0.04, -0.02, 0.3) rotation : (0, -0, 0) [ map_builder_options.cc: 103 ] [INFO] Load configurations from file: ./config/lidar_only.xml

configs from file

                  whole_options.export_file_path -> pcd/
                  whole_options.map_package_path -> pkgs/test/
                   whole_options.odom_calib_mode -> [default] 1
                   whole_options.separate_output -> [default] false
        whole_options.output_direct_combined_map -> true
                       whole_options.output_mrvm -> false
                 whole_options.output_kitti_pose -> [default] false
                     whole_options.separate_step -> [default] 200

             output_mrvm_settings.output_average -> false
                 output_mrvm_settings.output_rgb -> false
          output_mrvm_settings.use_max_intensity -> [default] true
             output_mrvm_settings.prob_threshold -> 0.57
             output_mrvm_settings.low_resolution -> 1
            output_mrvm_settings.high_resolution -> 0.1
                   output_mrvm_settings.hit_prob -> 0.55
                  output_mrvm_settings.miss_prob -> 0.48
                   output_mrvm_settings.z_offset -> 0
      output_mrvm_settings.max_point_num_in_cell -> 10

                          Creating filter: Range
                                      min_range_ -> 4.5
                                      max_range_ -> 75
                  Creating filter: RandomSampler
                                  sampling_rate_ -> 0.5

                                   options->type -> 6
                     options->accepted_min_score -> [default] 0.7
          front_end_options.accumulate_cloud_num -> 360
         motion_filter_options.translation_range -> 1
               motion_filter_options.angle_range -> 1.5
                motion_filter_options.time_range -> 0.25
              motion_compensation_options.enable -> true
         motion_compensation_options.use_average -> true
                             imu_options.enabled -> false
                                imu_options.type -> 0
                           imu_options.frequency -> 200
                    imu_options.gravity_constant -> 9.8
                                   options->type -> 6
                     options->accepted_min_score -> 0.7
                      submap_options.frame_count -> 2
                submap_options.enable_inner_mrvm -> [default] false
              submap_options.enable_voxel_filter -> false
          submap_options.enable_random_sampleing -> false
             submap_options.random_sampling_rate -> 0.5
                       submap_options.voxel_size -> [default] 0.1
               submap_options.enable_disk_saving -> true
                     submap_options.enable_check -> false
                submap_options.disk_saving_delay -> 30
               submap_options.saving_name_prefix -> s_
                 isam_optimizer_options.use_odom -> false
                  isam_optimizer_options.use_gps -> false
         isam_optimizer_options.output_graph_pic -> [default] false
   isam_optimizer_options.enable_extrinsic_calib -> [default] true
      isam_optimizer_options.gps_factor_init_num -> [default] 25
   isam_optimizer_options.gps_factor_sample_step -> [default] 1
isam_optimizer_options.gps_factor_init_angle_rad -> [default] 1.6
                   loop_detector_setting.use_gps -> false
            loop_detector_setting.use_descriptor -> true
      loop_detector_setting.output_matched_cloud -> false
  loop_detector_setting.trying_detect_loop_count -> 1
     loop_detector_setting.loop_ignore_threshold -> 20
   loop_detector_setting.nearest_history_pos_num -> 5
   loop_detector_setting.max_close_loop_distance -> 5
 loop_detector_setting.max_close_loop_z_distance -> 2
          loop_detector_setting.m2dp_match_score -> 0.975
   loop_detector_setting.accept_scan_match_score -> 0.8
                      map_package_options.enable -> false
               map_package_options.border_offset -> 100
                 map_package_options.piece_width -> 500
           map_package_options.cloud_file_prefix -> part_
           map_package_options.descript_filename -> map_package.xml

---

[ map_builder.cc: 63 ] [INFO] Init scan matchers. dist_outlier_ratio -> 0.7 knn_normal_estimate -> 7 max_iteration -> 100 [ map_builder.cc: 76 ] [INFO] Init isam optimizer. [ map_builder.cc: 218 ] [INFO] Add a new trajectory : 0 [ map_builder.cc: 102 ] [INFO] Init threads. [ map_builder.cc: 111 ] [INFO] Init finished. [ map_builder.cc: 783 ] [INFO] disable odom. [ map_builder.cc: 792 ] [INFO] disable gps. [ memory_manager.cc: 53 ] [INFO] Start managing memory. processed 8.042402 of 746.962851 seconds of the bag... ... processed 741.131710 of 746.962851 seconds of the bag... [ map_builder.cc: 716 ] [INFO] Finishing Remaining Computations... [ map_builder.cc: 396 ] [INFO] point cloud thread exit. [ map_builder.cc: 665 ] [INFO] no enough frames for new submap, quit [ map_builder.cc: 712 ] [INFO] submap processing done. [ map_builder.cc: 566 ] [INFO] All submaps have been connected. [ map_builder.cc: 616 ] [INFO] generating path files ... [ memory_manager.cc: 62 ] [INFO] End managing memory. block name | times | avg.(ms) | sum.(ms) | min.(ms) | max.(ms) main | 1 | 177011 | 177011 | 177011 | 177011

[EdwardLiuyc]

@lvjunning1243015876 I suppose you may have some misunderstanding on accumulate_cloud_num, this parameter means how many RAW point clouds you want to be merged as 1 point cloud. As you set it to 360 (if I get it right, you are using velodyne and the frequency of point cloud message is 10Hz), then every 36s you will get 1 point cloud to do mapping, which is kind of weird. I suggest that you try with accumulate_cloud_num set to 1.

[LvJohny]

Thanks very much. But when I set accumulate_cloud_num to 1, it still informs no enough frames for new submap, quit.

[EdwardLiuyc]

My guess is that you set the wrong frame_id or topic_name for lidar, or the input filter removed most of the points. Could you please share me the rosbag you use so I can also check? (Through google drive or some other cloud drive).

[LvJohny]

Yes, this is google driver link https://drive.google.com/file/d/1YSiRurMOfZEVrI21AmGO0ZRYU4bCKBvK/view?usp=sharing.

[EdwardLiuyc]

@lvjunning1243015876 I tried with your data but unfortunately I didn't got a good result from it. Regarding your problem, actually the POINT_CLOUD_TOPIC should be set to /velodyne_points.

As your use case, I am not very confident if you should use this project. First is that this project is mainly tested with dataset collected by vehicles, so roll and pitch angle are always changing slowly. As far as I know, UAV senario is not friendly to Lidar SLAM becuase the high angle velocity will cause large distortion on raw point cloud which is really hard for motion compensation. For that, I also tested with imu data in your rosbag. The result is still a little bit disappointing. Secondly, your raw data shows really little difference in time. See images below. This is like a hallway problem. Wrong match happens a lot.

My suggestion (if you are still interested in this project):

1. Test your setup in a more feature-rich senario.
2. Or design a feature-rich path for data collection.
3. Achieve hardward level synchronization with lidar and imu.

Still, I will keep investigating with your data. :blush:

[LvJohny]

Thank you for your patience and suggestions very much. And I will try other way to map.