[Phd] Project proposals from today's IFRoS meeting

[Pál Márton]

Dear Colleagues and Students,

Thanks to those who participated in the meeting.
I am sharing the proposals that we discussed. If any of these raise a question or interest in you, do not hesitate to contact us!
Consider these as initiatives from which BSc/MSc theses and PhD subprojects can root.

Best regards,
Márton

1. Reliability assessment of high-precision (RTK) GNSS-based waypoint navigation – comparison between urban and field environments
In this project, the student examines how the accuracy and stability of RTK GNSS affect waypoint-based robot navigation under different environmental conditions (open field, forested area, built-up area). During the robot’s movement, GNSS data must be fused with IMU and wheel odometry data. The ideal outcome is centimeter-level route tracking accuracy and a statistical evaluation of the error distribution. Scenarios should also be formulated for future error correction.
2. Real-time 3D modeling through LiDAR and RGB camera data fusion
The goal of this project is to build a system that combines Velodyne LiDAR point clouds with camera images to construct a real-time 3D model of the robot’s immediate surroundings. Processing should take place in a ROS 2 environment, relying on SLAM algorithms. The resulting 3D model should then be analyzed for its usability in obstacle detection and avoidance between waypoints.
3. GNSS-free navigation through integration of open-access map databases
The robot must navigate based on open map databases (e.g., OpenStreetMap) without using GNSS. This requires identifying a data format that can be efficiently integrated into the robot’s system. The goal is to examine how the robot can compare data from its onboard sensors (LiDAR, camera) with map layers. Using a SLAM algorithm, the robot constructs its own spatial database, and when it detects discrepancies or additional objects compared to the reference data, it updates its internal map. It is worth assessing how accurately the robot can correct the map database with the available tools, and how quickly and reliably it detects and handles changes.
4. Comparison of RTK GNSS and SLAM algorithms in robotic navigation
The aim of this thesis (or potentially an extended research project) is to compare two different positioning and navigation techniques — RTK GNSS and LiDAR-camera-based SLAM — during a robot’s operation and spatial data collection. The task involves integrating both methods into one system, combining the global accuracy of GNSS with the local detail richness of a SLAM-generated virtual environment. To evaluate performance, test environments of varying physical characteristics and complexity must be defined, where navigation accuracy and reliability can be measured.
5. Waypoint-based route planning with obstacle and traversability assessment - direct contact: Prof. Gáspár Albert (albert at ludens.elte.hu)
The project aims to develop a route-planning system that not only targets reaching predefined waypoints but also dynamically evaluates traversability (e.g., obstacles, slope, surface type) based on sensor data. The robot decides in real time whether the planned path is passable and, if necessary, selects an alternative route. The investigation may cover various terrain types (asphalt, grass, gravel).
6. Detection of environmental changes and map database updating by robot - direct contact: Prof. Gáspár Albert (albert at ludens.elte.hu)
The robot’s goal is to map a sample area and then identify changes during repeated traversals. Using camera and LiDAR sensors, it detects new objects (e.g., constructions, vegetation, temporary obstacles) and updates the background database accordingly. The objective is to evaluate how independently the robot can keep a geospatial database up to date.

--------- következő rész ---------
Egy csatolt HTML állomány át lett konvertálva...
URL: <http://lazarus.elte.hu/pipermail/phd/attachments/20251201/426ebd47/attachment.html>