Skip to content

README.md

Latest commit

 

History

History
87 lines (65 loc) · 3.18 KB

README.md

File metadata and controls

87 lines (65 loc) · 3.18 KB

LiDAR Parsing with g3dlidarparse Element

This directory contains a script demonstrating how to use the g3dlidarparse element for parsing LiDAR point‑cloud frames.

The g3dlidarparse element ingests raw LiDAR frames (BIN/PCD) and attaches LidarMeta (points, frame_id, timestamps, stream_id) for downstream fusion, analytics, or visualization.

How It Works

The sample uses the GStreamer command‑line tool gst-launch-1.0 to build a pipeline string. Elements are separated by !, and properties are provided as property=value pairs.

This sample builds a pipeline of:

  • multifilesrc for reading sequential LiDAR binary files
  • g3dlidarparse for parsing LiDAR frames and attaching metadata
  • fakesink for discarding output (metadata is attached to buffers)

The g3dlidarparse element performs:

  1. Input parsing: Reads raw LiDAR frames from application/octet-stream
  2. Frame thinning: Applies stride and frame-rate controls
  3. Metadata attachment: Emits LidarMeta with point cloud data

Prerequisites

1. Verify DL Streamer Installation

Ensure DL Streamer is properly compiled and the g3dlidarparse element is available:

gst-inspect-1.0 g3dlidarparse

If the element is found, you should see detailed information about the element, its properties, and pad templates.

2. Download Lidar Data and Configuration

Download the sample lidar binary dataset:

DATA_DIR=velodyne
echo "Downloading sample LiDAR frames to ${DATA_DIR}..."
TMP_DIR=$(mktemp -d)
git clone --depth 1 --filter=blob:none --sparse https://github.com/open-edge-platform/edge-ai-suites.git "${TMP_DIR}/edge-ai-suites"
pushd "${TMP_DIR}/edge-ai-suites" >/dev/null
git sparse-checkout set metro-ai-suite/sensor-fusion-for-traffic-management/ai_inference/test/demo/kitti360/velodyne
popd >/dev/null
mkdir -p "${DATA_DIR}"
cp -a "${TMP_DIR}/edge-ai-suites/metro-ai-suite/sensor-fusion-for-traffic-management/ai_inference/test/demo/kitti360/velodyne"/* "${DATA_DIR}/"
rm -rf "${TMP_DIR}"

This will create a velodyne directory containing the binary files of the lidar data.

Environment Variables

export GST_DEBUG=g3dlidarparse:5

Running the Sample

Usage:

./g3dlidarparse.sh [LOCATION] [START_INDEX] [STRIDE] [FRAME_RATE]

or

GST_DEBUG=g3dlidarparse:5 gst-launch-1.0 multifilesrc location="velodyne/%06d.bin" start-index=260 caps=application/octet-stream ! g3dlidarparse stride=5 frame-rate=5 ! fakesink

or

GST_DEBUG=g3dlidarparse:5 gst-launch-1.0 multifilesrc location="pcd/%06d.pcd" start-index=1 caps=application/octet-stream ! g3dlidarparse stride=5 frame-rate=5 ! fakesink

Where:

  • location points to a sequence of .bin or .pcd files (zero-padded index)
  • start-index selects the starting frame index
  • stride controls how often frames are processed
  • frame-rate throttles the output frame rate

Sample Output

The sample:

  • prints the full gst-launch-1.0 command to the console
  • outputs LiDAR parser debug logs and metadata summaries

See also