Autonomous Indoor Vehicle (AIV) Navigation Stack

B.Eng Final Year Project | Bells University of Technology

This repository contains the source code and ROS 2 configuration for the Design and Implementation of an Autonomous Indoor Vehicle, The project implements LiDAR-based Simultaneous Localization and Mapping (SLAM) and Frontier Exploration on a differential-drive mobile robot. It features a custom navigation stack built on ROS 2 Humble, capable of mapping indoor environments.

---

🤖 System Overview

The system is built upon the Waveshare WAVE ROVER chassis, utilizing a Raspberry Pi 4B as the central processing unit. It bridges high-level path planning with low-level motor control via a custom serial interface to an ESP32 driver board.

Key Features

• LiDAR-Based SLAM: Real-time 2D mapping and localization using the slam_toolbox and LD19 ToF LiDAR.
• Frontier Exploration: Autonomous navigation of unknown map frontiers to complete mapping of an environment.
• Custom Hardware Interface: A Python-based ROS 2 hardware abstraction layer converting geometry_msgs/Twist commands into serial packets for the Waveshare chassis.
• Sensor Integration: Fusion of LiDAR and IMU data for robust odometry.

---

🛠️ Hardware Architecture

• Chassis: Waveshare WAVE ROVER (4WD Metal Body, Differential Drive).
• Onboard Computer: Raspberry Pi 4B (4GB RAM) running Ubuntu 22.04 Server.
• LiDAR: LD19 D-ToF (12m range, 4500Hz sampling).
• Vision: A200 RGB-D Camera (used for depth estimation benchmarks).
• Low-Level Control: ESP32-WROOM-32 Driver Board (handles Motor PID and 9-axis IMU data).

---

📊 Research Findings & Benchmarks

As part of the thesis, this platform was used to benchmark and analyze SLAM performance across varying indoor conditions using a novel "Clutter Index" to quantify environmental complexity. The primary contribution was the documentation of performance degradation rather than a coded solution for it.

• Trajectory Accuracy (RMSE): 8.24 cm.
• Localization Accuracy: $\sim$10 cm relative to ground truth.
• Average Pose Error: 4.78 degrees (Orientation).
• Key Finding: Identified a linear correlation between environmental clutter levels and LiDAR point cloud noise standard deviation.

---

📦 Dependencies & Setup

Prerequisites

• OS: Ubuntu 22.04 LTS.
• Middleware: ROS 2 Humble Hawksbill.

Installation

# Clone the repository
git clone [https://github.com/doabayomi/aiv_project.git](https://github.com/doabayomi/aiv_project.git)

# Install dependencies
rosdep install --from-paths src --ignore-src -r -y

# Build the workspace
colcon build --symlink-install

🚀 Usage

This stack supports three primary operational modes via the wave_rover package:

1. System Initialization (Driver & Sensor Launch) This launches the LD19 LiDAR driver, IMU node, and the custom serial driver node for the Waveshare chassis.

ros2 launch wave_rover wave_rover

2. Manual Control (Teleoperation) After initialization, use a keyboard or game controller to manually teleoperate the vehicle for mapping or debugging.

ros2 run teleop_twist_keyboard teleop_twist_keyboard # (Standard ROS2 teleop node)

3. Autonomous Frontier Exploration Starts the SLAM process and launches the autonomous exploration behavior, directing the robot toward unknown frontiers.

ros2 launch wave_rover frontier_exploration# (Assuming this is the launch file name)

---

👤 Authors

Daniel Abayomi & Richard Okpara B.Eng Mechatronics Engineering