How to use PointCloud from pedsim_sensors to update costmap obstacles

edit

Hi, all

I have been trying to navigate pedsim_ros built-in diff-robot in environment created pedsim_similator

I have been able to navigate robot to desirable goal by navigation stack:

• I have recorded and placed at https://www.youtube.com/watch?v=d6atQ...

But as you seen, robot can't avoid obstacles (men) because I think that local costmap didn't update obstacles To update obstacles to local costmap, I have done as following:

• In local_costmap_params.yaml, I place two layers obstacle_layer and inflation_layer, obstacle_layer updated from _pedsim_people_sensor/point_cloud_local_ topic:

  plugins:

   - {name: obstacle_layer,          type: "costmap_2d::ObstacleLayer"}
  
   - {name: inflation_layer,         type: "costmap_2d::InflationLayer"}
  

  obstacle_layer:

   observation_sources: bump
  
   bump: {data_type: PointCloud , sensor_frame: odom, topic: pedsim_people_sensor/point_cloud_local, marking: true, clearing: true, min_obstacle_height: 0.0, max_obstacle_height: 3 }
  

Hint: Pls view full source code at: https://github.com/AMRobots/pedsim_teb

• I checked _pedsim_people_sensor/point_cloud_local_ topic, it is ok and can view it in rviz

--

I already have tried to solve this issue by:

• Method 1: set min_obstacle_height to 0.0 cm as suggestion at here

• Method 2: convert _pedsim_people_sensor/point_cloud_local_(PointCloud) to PointCloud2 (by point_cloud_converter package), then convert PointCloud2 to LaserScan (by pointcloud_to_laserscan package), finally use LaserScan to update local costmap

With two methods I failed.

pls suggest me the way to fix it.