I ran a simulation in Gazebo to try SLAM, and it worked great with a lidar with a FoV of 360°. However, after I reduced the angle to fit the lidar available to me, which has a FoV of 70°, the performance degraded significantly. Here (#239), you recommended reducing the values of the parameters _minimum_traveldistance and _minimum_travelheading in the configuration file and using an environment with many instances to improve performance. I reduced the value of the parameters to the values you recommneded and much lower, but the performance wasn't as good as before. As for the world, I used the Turtlebot house, which, I suppose, fulfills the world requirements. I made a round around the house, and the results are really bad. The world was mapped perfectly with the 360° beam, but the results with the 70° beam are disappointing:
This is the world that should be mapped. I walked around the table in the main room, then went to the other room to the right, circled the table there, and returned to the initial position:
I would be grateful for your help and any recommendations on how to improve performance.
I ran a simulation in Gazebo to try SLAM, and it worked great with a lidar with a FoV of 360°. However, after I reduced the angle to fit the lidar available to me, which has a FoV of 70°, the performance degraded significantly. Here (#239), you recommended reducing the values of the parameters _minimum_traveldistance and _minimum_travelheading in the configuration file and using an environment with many instances to improve performance. I reduced the value of the parameters to the values you recommneded and much lower, but the performance wasn't as good as before. As for the world, I used the Turtlebot house, which, I suppose, fulfills the world requirements. I made a round around the house, and the results are really bad. The world was mapped perfectly with the 360° beam, but the results with the 70° beam are disappointing:
This is the world that should be mapped. I walked around the table in the main room, then went to the other room to the right, circled the table there, and returned to the initial position:
I would be grateful for your help and any recommendations on how to improve performance.
This is my urdf file:
This is my launch file slam_algorithm.launch.py:
This is the configuration file mapper_params_online_async.yaml:
for reproduction, this is the rviz configuration file urdf_config.rviz: