Weedbot Perception Node (ESP32 + micro-ROS)

This repository contains the perception firmware from my weeding robot dissertation.
It demonstrates a lightweight way to integrate a 2D LiDAR + IMU into a ROS 2 architecture using only:

• a microcontroller (ESP32), and
• a computer running ROS 2 (laptop/desktop)

…without needing a higher-level onboard computer (e.g., Raspberry Pi) to interface sensors directly.

✅ Outputs to ROS 2:

• scan (sensor_msgs/LaserScan)
• imu/data (sensor_msgs/Imu)

Related repos (same dissertation project):

• Delta / End Effector node: https://github.com/busayojee/weedbotDeltaNode
• Drive base node: https://github.com/busayojee/weedbotDriveNode

---

Overall System Architecture

What this repo is for

If you’re building a robot with ROS2 and you need to bring sensors into ROS2 but only have a microcontroller (ESP32) near the sensors, this pattern works well:

ESP32 (sensor I/O) → micro-ROS (XRCE-DDS) → Wi‑Fi UDP → ROS 2 host

The ESP32 handles:

• UART + motor control for RPLIDAR A1
• I2C for BNO08x IMU
• Packaging data into ROS 2 message types (LaserScan, Imu)
• Streaming to the ROS 2 host through micro_ros_agent

Your computer handles:

• SLAM / localization / fusion / navigation (e.g., SLAM Toolbox, Nav2, robot_localization)

---

High-level architecture

Data flow

1. ESP32 reads LiDAR over UART and batches scan samples.
2. ESP32 reads IMU over I2C (rotation vector + gyro).
3. ESP32 publishes ROS 2 messages via micro-ROS over Wi‑Fi UDP.
4. A ROS 2 computer runs micro_ros_agent and receives the topics.

This keeps sensor wiring short and lets you place sensors on a robot platform without running USB cables back to a laptop.

---

Hardware

• ESP32 (tested with PlatformIO esp32dev)
• RPLIDAR A1 (UART + motor enable pin)
• Adafruit BNO08x IMU (I2C)

---

ROS 2 interface

Node

• Node name: esp32_lidar_imu
• Namespace: "" (empty)

Topics

• scan → sensor_msgs/LaserScan
• imu/data → sensor_msgs/Imu

Frame IDs

• IMU: imu_link
• LiDAR: laser_frame

---

Wiring / Pins (from include/weedbot_config.h- configurable)

RPLidar (UART + motor)

• LIDAR_RX_PIN = 21
• LIDAR_TX_PIN = 22
• LIDAR_MOTOR_PIN = 2

IMU (I2C)

• I2C_SDA_PIN = 19
• I2C_SCL_PIN = 18

---

Repo structure

src/
  main.cpp                     # entry point
lib/
  node/                        # app orchestration
  ros/                         # micro-ROS context (node, pubs, timer, executor)
  sensors/                     # BNO08x + LiDAR wrappers (More sensors could be added)
include/
  weedbot_config.h             # user-editable config

---

PlatformIO dependencies

Copy/paste into platformio.ini:

[env:esp32dev]
platform = espressif32
board = esp32dev
framework = arduino
build_flags = -I include
monitor_speed = 115200
board_microros_transport = wifi
lib_deps =
    https://github.com/micro-ROS/micro_ros_platformio
    mirs240x/micro_ros_arduino@^2.0.7-humble
    https://github.com/sjamthe/ESP32RPLidar.git
    https://github.com/micro-ROS/micro_ros_arduino.git
    adafruit/Adafruit BusIO@^1.17.1
    hideakitai/MPU9250@^0.4.8
    adafruit/Adafruit BNO08x@^1.2.5

---

Configuration

Edit: include/weedbot_config.h

Key values:

• Wi‑Fi: WIFI_SSID, WIFI_PWD
• Agent: AGENT_IP, AGENT_PORT
• Topics: TOPIC_SCAN, TOPIC_IMU
• Frames: LASER_FRAME_ID, IMU_FRAME_ID
• IMU publish period: IMU_PERIOD_MS = 100 (10 Hz)
• BNO08x report intervals:
  • BNO_ROTATION_VECTOR_US = 5000 (200 Hz)
  • BNO_GYRO_CAL_US = 10000 (100 Hz)
• LiDAR batching:
  • LIDAR_SCANS_PER_PUBLISH = 5
  • MAX_BATCH_MEASUREMENTS = 1200
• Agent ping: AGENT_PING_EVERY_MS = 20000

---

Build & flash (PlatformIO)

pio run -t upload

---

Run on ROS 2 (host computer)

Start the micro-ROS agent:

ros2 run micro_ros_agent micro_ros_agent udp4 --port 8888

Verify topics:

ros2 topic list
ros2 topic echo /scan
ros2 topic echo /imu/data

---

Reuse / adapt for your robot

You can adapt this repo if you:

• want to integrate LiDAR + IMU into ROS 2 using only a microcontroller + PC,
• need a pattern for micro-ROS sensor bridging, or
• want to extend the node with additional sensors (GPS, encoders, ToF, cameras via separate nodes, etc.).

Typical follow-on integrations:

• robot_localization (EKF) consuming /imu/data
• SLAM Toolbox consuming /scan
• TF tree setup linking laser_frame and imu_link into your base frames

---

Notes on Time Sync

If you add timestamps or additional ROS messages later, keep the time sync behavior from the perception stack. The agent time ensures consistent stamps for logging and coordination.

Credits and References

This perception firmware is part of the dissertation project Control and Navigation of a Weeding Robot, together with the four wheel differential drive, removal, and higher level autonomy. If you have any question regarding this driver please do not hesitate to reach out.