Abstract
This project involved building and deploying an Autonomous Unmanned Ground Vehicle (UGV) from the ground up, combining mechanical design, electronics, control systems, and advanced autonomy software. I led the software team, designed the communication architecture, and implemented ROS-based controllers, localization, mapping, and navigation pipelines. The UGV was designed to operate in open-world environments, collect data wirelessly, and autonomously explore unknown areas using probabilistic mapping and motion planning.
Introduction
UGVs are critical for real-world applications ranging from search-and-rescue to defense, agriculture, and exploration. Unlike lab environments, real-world terrains demand robust localization, navigation, and communication architectures that can withstand uncertainty and noise. This project aimed to design such a platform by combining custom hardware integration with cutting-edge autonomy stacks like LioSAM, AMCL, and RRT-based exploration.
Motivation
Build a UGV capable of autonomous navigation in complex and unstructured environments.
Integrate modern sensors like Livox Mid-360 LiDAR for accurate 3D mapping.
Design a modular architecture for ROS-based control and data transmission.
Enable autonomous exploration of unknown terrains, making the robot scalable to industrial and research applications.
System Overview
Custom ROS-Based Controller
Designed and implemented a speed-controlled ROS driver for ODrive motor controllers.
Created a proprietary data transmission pipeline for ODrive feedback integrated into ROS.
SLAM & Localization
Adaptive Monte Carlo Localization (AMCL) for pose estimation.
Particle filtering fused Livox Mid-360 LiDAR with odometry feedback from ODrive.
LioSAM Mapping with Livox data for accurate 3D LiDAR SLAM.
Built a probabilistic occupancy grid (OctoMap) for navigation and exploration.
Autonomous Navigation
Implemented RRT-based global and local navigation.
Integrated the Autonomous Exploration Planner to allow the UGV to actively explore open-world environments.
Mechanical & Hardware Contributions
Assisted in mechanical assembly and machining parts of the UGV.
Performed power supply calculations and ordered custom supply modules and connectors.
Designed the electronics architecture to reliably power motors, sensors, and compute.
Communication & Data Pipeline
Developed a wireless data collection and transmission system for remote monitoring.
Designed the communication architecture for real-time ROS integration between ODrive, sensors, and the onboard compute.
Leadership Role
Led the software team for the UGV project.
Designed the overall control system architecture and ensured robust integration across hardware and software layers.
Pipeline Diagram
Odrive ROS Controller → Speed control + odometry feedback
Sensor Suite (Livox LiDAR, IMU, Odometry) → Localization + SLAM (AMCL, LioSAM, OctoMap)
Autonomous Planner → RRT-based navigation + Exploration Planner
ROS Communication Layer → Synchronizes data & commands
UGV Hardware → Actuation + feedback + wireless telemetry
Results
Built a stable ROS driver for ODrive enabling smooth speed control.
Achieved accurate localization using AMCL and particle filter fusion.
Produced dense 3D maps with LioSAM and OctoMap.
Enabled autonomous navigation and exploration in real-world test environments.
Delivered a modular software stack and reliable communication architecture for scaling future UGV projects.
Conclusion
The Autonomous UGV project was a full-stack challenge — from machining aluminum brackets to implementing SLAM pipelines and writing ROS drivers. By leading the software team, I learned how to bridge hardware and autonomy into a single working system. The end result was more than just a mobile robot: it was a scalable, autonomous platform capable of exploring unknown environments and setting the foundation for future robotics research and field deployment.
