Accumulated positioning error
Wheel-based odometry drifts over time. QR landmarks were introduced as known reference points to periodically correct the estimated pose.
Robotics · Autonomous Systems
Autonomous AGV Mapping System
A LiDAR-based mapping and localization system for an autonomous industrial guided vehicle.
Year
2024–2026
Status
Prototype
Role
Mapping Software, Sensor Processing, Interface Integration
Duration
Competition & Independent Development
Context and objective
This project focuses on real-time mapping, localization and navigation for an industrial AGV. It combines 2D LiDAR scans, wheel odometry, IMU data and QR-based absolute position corrections to build a more reliable understanding of the vehicle's environment.
The Problem
What needed to change?
Autonomous vehicles operating indoors must estimate their position despite wheel slip, sensor noise and changing surroundings. Odometry alone accumulates error, while raw LiDAR data requires alignment before it can be added to a consistent map.
The Solution
How the system responds
I implemented a mapping workflow that converts polar LiDAR measurements into Cartesian points, filters noise, aligns scans with ICP and updates the map over time. Encoder and IMU data support motion estimation, while QR markers provide absolute correction points.
Key features
✓360-degree 2D LiDAR data processing
✓Polar-to-Cartesian coordinate conversion
✓ICP scan matching with KDTree and SVD
✓Encoder and IMU-assisted odometry
✓QR-based absolute position correction
✓Obstacle detection and map maintenance
✓Dijkstra-based path planning
✓Real-time robot monitoring interface
Role and responsibilities
01
Developing the LiDAR mapping pipeline
02
Implementing ICP-based scan alignment
03
Combining odometry, IMU and QR positioning
04
Preparing live map visualization and simulation tools
05
Integrating robot data through WebSocket communication
06
Supporting autonomous route planning and mission monitoring
Tools used in the project
PythonRPLIDARRaspberry PiWebSocketROS 2NumPy
Challenges and solutions
Noisy and moving map points
Raw scans may include temporary obstacles and measurement noise. Filtering and point-age logic help keep the map stable and remove stale points.
Results and impact
Built a real-time LiDAR mapping workflow
Combined relative and absolute localization methods
Created simulation tools for recorded sensor data
Integrated mapping outputs with navigation and monitoring interfaces
Selected screens and system views
Need a similar solution?
Related projects
Enterprise Software · Decision Support
BOTAŞ Workforce Planning System
A role-based workforce planning platform designed to improve employee visibility, skill matching and operational task assignment.
Artificial Intelligence · Data ScienceRational Medicine Optimization System
A data-driven research project focused on medication waste, prescription planning, healthcare risk analysis and decision support.
Mobile Application · Artificial IntelligenceAI-First Planning Assistant
An AI-assisted planning application that turns natural-language requests into personal and collaborative plans.
Have a similar project in mind?
Let’s discuss the workflow, technical requirements and product direction behind your idea.
Discuss Your Project