fss_sensing

This package provides a ROS 2 local point cloud simulator based on marsim_render. It renders a LiDAR point cloud from the current drone pose using a static PCD map.

The simulator supports the fss_time coordinated simulation clock and lock-step execution. The coordinated clock is enabled by default (use_fss_sim_time:=true), so the sensing timer advances with the global simulation time rather than wall time.

# Launch the local point cloud simulator with the default configuration.
ros2 launch fss_sensing fss_local_pointcloud_sim.launch.py

# Run with wall time instead of the FastSwarmSim coordinated simulation clock.
ros2 launch fss_sensing fss_local_pointcloud_sim.launch.py use_fss_sim_time:=false

# Run with a custom point cloud configuration file.
ros2 launch fss_sensing fss_local_pointcloud_sim.launch.py config_path:=/path/to/local_pointcloud_sim.yaml

What it contains:

A local point cloud simulator node that subscribes to the drone pose or odometry and renders the point cloud visible from the drone using marsim_render.
A static global point cloud publisher. The global_pc topic uses reliable, transient-local QoS, so subscribers that start after the simulator still receive the map.
A configurable LiDAR model and map source. The default configuration is installed at config/local_pointcloud_sim.yaml and supports the sensing range, rate, field of view, angular resolution, downsampling resolution, input topics, and output topics.
fss_time lock-step simulation-time integration. Set use_fss_sim_time:=false to use the normal ROS clock (/clock when use_sim_time is enabled, otherwise wall time).

Topics

The default configuration uses the following topics:

Topic	Type	Description
`mavros/local_position/pose`	`geometry_msgs/msg/PoseStamped`	Input drone pose. Set `use_odom: true` to use `mavros/local_position/odom` instead.
`cloud_registered`	`sensor_msgs/msg/PointCloud2`	Local point cloud rendered at the configured sensing rate (10 Hz by default).
`global_pc`	`sensor_msgs/msg/PointCloud2`	Complete static map point cloud, published once at startup.

Topic names, frame ID, map, and LiDAR parameters can be changed in local_pointcloud_sim.yaml.

Trouble shooting

Local point cloud is only published at 1 Hz

The default local point cloud rate is 10 Hz. It can also run at 2x real-time simulation speed. If cloud_registered is received at only about 1 Hz, the rendered point cloud is likely larger than the default DDS shared-memory segment or UDP receive buffer. Configure the active RMW implementation before launching ROS 2 nodes.

Fast DDS (`rmw_fastrtps_cpp`)

Fast DDS uses shared memory for local communication, but its default 256 KB segment is too small for large point clouds. Create fastdds.xml with a larger segment; 8 MB is a suitable starting point:

<?xml version="1.0" encoding="UTF-8"?>
<dds xmlns="http://www.eprosima.com">
  <profiles>
    <transport_descriptors>
      <transport_descriptor>
        <transport_id>shm_transport_8MB</transport_id>
        <type>SHM</type>
        <segment_size>8388608</segment_size>
      </transport_descriptor>
    </transport_descriptors>
    <participant profile_name="default_participant" is_default_profile="true">
      <rtps>
        <userTransports>
          <transport_id>shm_transport_8MB</transport_id>
        </userTransports>
        <useBuiltinTransports>true</useBuiltinTransports>
      </rtps>
    </participant>
  </profiles>
</dds>

Then start ROS 2 with the profile enabled:

export RMW_IMPLEMENTATION=rmw_fastrtps_cpp
export FASTDDS_DEFAULT_PROFILES_FILE=/path/to/fastdds.xml
# Required by older ROS 2 Humble releases; harmless to set as well.
export FASTRTPS_DEFAULT_PROFILES_FILE=/path/to/fastdds.xml
ros2 daemon stop

Set segment_size according to the largest point cloud message. Every DDS participant (normally one per process) allocates a segment, so avoid setting it unnecessarily large.

Cyclone DDS (`rmw_cyclonedds_cpp`)

Cyclone DDS uses UDP by default. Large local point clouds can overflow the system UDP receive buffer, so increase it before launching the nodes:

sudo tee /etc/sysctl.d/60-cyclonedds.conf >/dev/null <<'EOF'
net.core.rmem_max=8388608
net.core.rmem_default=8388608
EOF
sudo sysctl --system

export RMW_IMPLEMENTATION=rmw_cyclonedds_cpp
ros2 daemon stop

The 8 MB values are a starting point and should be increased only when the largest point cloud requires it. Cyclone DDS shared-memory transport requires additional RouDi/mempool configuration and is not recommended here; use the UDP configuration above.

See the large point cloud DDS configuration reference for background and further tuning considerations.