Skip to content

Sensor_Fusion_Guide.md

Latest commit

 

History

History
103 lines (83 loc) · 3.48 KB

Sensor_Fusion_Guide.md

File metadata and controls

103 lines (83 loc) · 3.48 KB

Sensor Fusion Guide: LiDAR + Ultrasonic Integration

This document explains the sensor fusion methodology for combining TF-Luna LiDAR and HC-SR04 ultrasonic data in our autonomous vehicle.

Fusion Architecture Overview

graph LR
    A[TF-Luna LiDAR] -->|I²C @ 100Hz| C[Fusion Processor]
    B[HC-SR04 Ultrasonic] -->|GPIO @ 20Hz| C
    C --> D[Kalman Filter]
    D --> E[Confidence-Weighted Fusion]
    E --> F[Navigation Controller]
Loading

Key Parameters

Sensor Range Precision Update Rate Best Use Case
TF-Luna LiDAR 0.1m - 8.0m ±3cm 100Hz Long-range mapping
HC-SR04 0.02m - 4m ±0.5cm 20Hz Close-range safety

Fusion Algorithm

Step 1: Coordinate Transformation

// Transform LiDAR point cloud to vehicle coordinates
void transformLidarData() {
    for (int i = 0; i < lidar_points; i++) {
        float angle = lidar.getAngle(i);
        float distance = lidar.getDistance(i);
        x[i] = distance * cos(angle - LIDAR_MOUNT_OFFSET);
        y[i] = distance * sin(angle - LIDAR_MOUNT_OFFSET);
    }
}

Step 2: Kalman Filtering

Each sensor gets its own Kalman filter to reduce noise:

# Pseudocode for ultrasonic Kalman filter
Q = 0.01  # Process noise
R = 0.1   # Sensor noise

def kalman_update(z, prev_estimate, uncertainty):
    # Prediction
    pred_estimate = prev_estimate
    pred_uncertainty = uncertainty + Q
    
    # Update
    kalman_gain = pred_uncertainty / (pred_uncertainty + R)
    new_estimate = pred_estimate + kalman_gain * (z - pred_estimate)
    new_uncertainty = (1 - kalman_gain) * pred_uncertainty
    
    return new_estimate, new_uncertainty

Step 3: Confidence-Weighted Fusion

// Fuse sensor readings based on confidence
float fusedDistance() {
    float lidar_dist = getFilteredLidar();
    float us_dist = getFilteredUltrasonic();
    
    // Weighting factors (tunable)
    float lidar_confidence = 0.7 * (1 - exp(-0.5 * lidar_dist));
    float us_confidence = 0.9 * exp(-0.1 * us_dist);
    
    // Normalize
    float total = lidar_confidence + us_confidence;
    
    return (lidar_confidence * lidar_dist + us_confidence * us_dist) / total;
}

Calibration Procedure

  1. Static Calibration:

    Place target at known distances (0.5m, 1m, 2m, 4m)
Run calibration.ino to record sensor error
Update src/sensor_calibration.h with offset values

  2. Dynamic Calibration:

    while moving:
    if ultrasonic_change > threshold and lidar_change < threshold:
        adjust_ultrasonic_gain(0.95)
    elif lidar_change > threshold and ultrasonic_change < threshold:
        adjust_lidar_gain(1.05)

Troubleshooting Table

Symptom Likely Cause Solution
False emergency stops Ultrasonic multipath Add foam ring around sensor
LiDAR dropout in sunlight IR interference Use sun shield enclosure
Inconsistent fusion output Time synchronization Add hardware interrupt sync
Sensor conflict > 1m Calibration drift Re-run static calibration

Pro Tip: Always validate sensor readings in the Serial Plotter before autonomous operation!

Next: ROS Integration