LiDAR Handler

The LiDARHandler (src/handlers/LiDARHandler.h & LiDARHandler.cpp) manages runtime spatial queries against preprocessed LiDAR point cloud databases, providing real-time terrain topology and obstacle metrics to GeoPlanner and VisualizationHandler.

---

1. Primary Responsibilities

1. DuckDB Database Interfacing: Connects to pre-built DuckDB database files (constants::LIDAR_HANDLER_DB_PATH) containing millions of geospatial points derived from USGS 3DEP LAS 1.4 point clouds.
2. Radial Spatial Lookups: Executes fast spatial queries to extract terrain points within a specified radius $R$ of an Easting/Northing coordinate.
3. Multi-Property Filtering: Supports filtering points by classification (e.g., ground vs vegetation vs structures), surface normal vectors, terrain slope, roughness, and curvature.
4. Traversal Metrics Provisioning: Provides precomputed traversal scores to the GeoPlanner costmap generator.

---

2. Point Data Architecture

Points are stored and returned in the LiDARHandler::PointRow structure:

struct PointRow
{
    int nID;                         // Unique point identifier
    double dEasting;                 // UTM Easting coordinate (meters)
    double dNorthing;                // UTM Northing coordinate (meters)
    double dAltitude;                // Altitude above sea level (meters)
    std::string szZone;              // UTM Zone designator
    std::string szClassification;    // Point classification (ground, rock, etc.)
    double dNormalX;                 // X component of local surface normal vector
    double dNormalY;                 // Y component of local surface normal vector
    double dNormalZ;                 // Z component of local surface normal vector
    double dSlope;                   // Surface slope (degrees)
    double dRoughness;               // Local terrain roughness metric
    double dCurvature;               // Surface curvature metric
    double dTraversalScore;          // Composite score [0.0 = impassable, 1.0 = smooth]
};

Filtering via PointFilter

Queries can be conditioned using LiDARHandler::PointFilter, specifying min/max bounds on slope, roughness, curvature, and normal vectors to isolate specific terrain hazards.

---

3. Database Engine: DuckDB

The handler leverages DuckDB rather than traditional relational engines:

• Columnar Execution Engine: Optimized for analytical vectorized queries on large numerical datasets.
• Embedded Operation: Runs in-process without requiring background server daemons.
• Thread Safety: Uses std::shared_mutex to allow concurrent read queries across GeoPlanner and VisualizationHandler threads.

---

4. Usage Example

// Opening the database during initialization (in main.cpp)
globals::g_pLiDARHandler = new LiDARHandler();
if (!globals::g_pLiDARHandler->OpenDB(constants::LIDAR_HANDLER_DB_PATH))
{
    LOG_ERROR(logging::g_qSharedLogger, "Failed to open LiDAR DuckDB database.");
}

// Querying terrain within a 15-meter radius of the rover
std::vector<LiDARHandler::PointRow> vNearbyPoints;
vNearbyPoints = globals::g_pLiDARHandler->GetPointsInRadius(stRoverUTM, 15.0);

// Filtering for steep obstacles (slope > 25 degrees)
LiDARHandler::PointFilter stFilter;
stFilter.dEasting  = stRoverUTM.dEasting;
stFilter.dNorthing = stRoverUTM.dNorthing;
stFilter.dRadius   = 20.0;
stFilter.dSlope    = LiDARHandler::PointFilter::Range<double>{25.0, 90.0};

std::vector<LiDARHandler::PointRow> vSteepObstacles;
vSteepObstacles = globals::g_pLiDARHandler->GetPointsWithFilter(stFilter);

---

5. LiDAR Data Sources & Web Inspection Tools

A. USGS LiDAR Point Cloud Storage Repository

The spatial elevation and terrain point clouds queried by LiDARHandler are sourced from the USGS 3D Elevation Program (3DEP) and processed into indexed DuckDB databases. Raw LAS/LAZ point cloud tiles, pre-generated DuckDB database artifacts, and ingestion scripts are hosted on the team’s GitLab server:

• USGS LiDAR Dataset Repository: MissouriMRDT/USGS_Data
• MRDT GitLab Organization: MissouriMRDT GitLab

Developers running simulations or offline tests requiring local terrain maps should acquire the appropriate regional .duckdb tiles from USGS_Data and place them at the path configured in constants::LIDAR_HANDLER_DB_PATH (data/LiDAR/ by default).

B. Online LiDAR Visualizer Tool

Terrain point clouds, cross-sectional elevation profiles, and traversability slopes can be visualized interactively in the web browser without launching local DuckDB instances:

• Interactive LiDAR Visualizer: visualizer.themrdt.org/lidar-tool/

This tool supports inspecting 3D colored point distributions, evaluating elevation gradients, and testing traversability threshold configurations across competition terrains.