line_sensor_data_tools

Small tools to record Stretch line sensor ranges and replay them for filter tests.

Build

cd <your_ros2_ws>
colcon build --packages-select line_sensor_data_tools
source install/setup.bash

Record

ros2 run line_sensor_data_tools line_sensor_record

This saves JSONL files in the package data directory. The first line stores the line sensor metadata needed for offline replay.

Optional:

ros2 run line_sensor_data_tools line_sensor_record --rate 30 --out test.jsonl

Stop with Ctrl+C.

Replay

ros2 run line_sensor_data_tools line_sensor_replay_filter --filter none

Or pick a file:

ros2 run line_sensor_data_tools line_sensor_replay_filter --file test.jsonl --filter xy_cluster

Published topics:

/line_sensor_trial/raw_candidates/obstacle_points
/line_sensor_trial/raw_candidates/small_drop_points
/line_sensor_trial/baseline/obstacle_points
/line_sensor_trial/baseline/small_drop_points
/line_sensor_trial/trial/obstacle_points
/line_sensor_trial/trial/small_drop_points
/line_sensor_trial/trial/clusters
/line_sensor_trial/trial/cluster_debug

raw_candidates are projected/classified bins before the current bin-level gates.

baseline is the current stretch4_under_base_hazard bin-level LineSensorSource output.

trial uses shared calibration/projection/classification, then your selected trial filter.

xy_cluster groups candidates by XY distance, then applies the existing run-level noise checks to each cluster: spray, point-noise, and valid-run. It then applies the same bin temporal confirmation as baseline. It does not use contiguous-bin grouping.

/line_sensor_trial/trial/clusters shows each XY cluster as a marker with a label like C3 keep n=8 c=4 or C4 spray n=16 c=0.

/line_sensor_trial/trial/cluster_debug publishes JSON with each cluster id, class, reason, centroid, radius, confirmed count, and sensor/bin members.

It does not run the rolling hazard map.

Old range-only JSONL files are not replayable offline. Record again so the file includes its own line sensor geometry/config metadata.

Add A Filter

Edit line_sensor_data_tools/replay_filter.py.

Add a function:

def filter_my_filter(source, status, state):
    candidates = _extract_candidates(source, status)
    return _hits_from_candidates(candidates)

Then add it:

FILTERS = {
    'none': filter_none,
    'moving_average': filter_moving_average,
    'xy_cluster': filter_xy_cluster,
    'my_filter': filter_my_filter,
}

Run it:

ros2 run line_sensor_data_tools line_sensor_replay_filter --filter my_filter