Tutorial: LiDAR Sensor Node (C++)

In this tutorial, you'll build a complete obstacle avoidance robot controller in C++. You'll learn:

  • Creating a scheduler with real-time settings
  • Publishing sensor data with the zero-copy loan pattern
  • Subscribing to data and processing it
  • Multi-node pipelines with execution ordering

Step 1: Create the Project

horus new lidar_robot --cpp
cd lidar_robot

Step 2: Write the Controller

Replace src/main.cpp with:

#include <horus/horus.hpp>
using namespace horus::literals;

int main() {
    horus::Scheduler sched;
    sched.tick_rate(100_hz).prefer_rt();

    // ── Topics ─────────────────────────────────────────────
    auto scan_pub = sched.advertise<horus::msg::LaserScan>("lidar.scan");
    auto scan_sub = sched.subscribe<horus::msg::LaserScan>("lidar.scan");
    auto cmd_pub  = sched.advertise<horus::msg::CmdVel>("cmd_vel");

    // ── Node 1: Simulated LiDAR Driver ─────────────────────
    sched.add("lidar_driver")
        .rate(100_hz)
        .order(0)
        .tick([&] {
            auto scan = scan_pub.loan();
            // Simulate: obstacles at 45° and 315°
            for (int i = 0; i < 360; ++i) {
                if (i > 40 && i < 50) {
                    scan->ranges[i] = 0.3f;  // close obstacle
                } else {
                    scan->ranges[i] = 5.0f;  // clear
                }
            }
            scan->angle_min = 0.0f;
            scan->angle_max = 6.28318f;
            scan_pub.publish(std::move(scan));
        })
        .build();

    // ── Node 2: Obstacle Avoidance Controller ──────────────
    sched.add("controller")
        .rate(50_hz)
        .order(10)
        .budget(5_ms)
        .on_miss(horus::Miss::Skip)
        .tick([&] {
            auto scan = scan_sub.recv();
            if (!scan) return;

            // Find minimum range in front 60° arc
            float min_front = 999.0f;
            for (int i = 150; i < 210; ++i) {
                if (scan->ranges[i] > 0.01f && scan->ranges[i] < min_front) {
                    min_front = scan->ranges[i];
                }
            }

            auto cmd = cmd_pub.loan();
            if (min_front < 0.5f) {
                cmd->linear  = 0.0f;
                cmd->angular = 0.5f;  // turn
            } else {
                cmd->linear  = 0.3f;
                cmd->angular = 0.0f;  // go straight
            }
            cmd_pub.publish(std::move(cmd));
        })
        .build();

    sched.spin();
}

Step 3: Build and Run

horus build
horus run

Step 4: Monitor

In another terminal:

horus topic list          # see active topics
horus topic hz cmd_vel    # check publish rate
horus node list           # see running nodes
horus topic echo cmd_vel  # watch velocity commands

Key Takeaways

  1. Captured pub/sub — create topics outside the tick lambda, capture by reference
  2. Loan pattern — pub.loan() gives you a direct SHM pointer (0ns data access)
  3. Execution order — order(0) runs before order(10), ensuring sensor data is fresh
  4. Budget enforcement — budget(5_ms) + on_miss(Skip) keeps the system responsive
  5. Zero configuration — no XML launch files, no msg files, no codegen step

Next Steps

  • Add a third node for motor control
  • Use horus record to capture and replay the session
  • Try horus sim3d for 3D simulation with physics