Active topics for student theses and indivudual/team projects.
We will provide all required access, hardware and sensors. We also have a lab environment with all lab equipment where you can perform development and experiments.
If you are a bachelor student and feel that you can handle some specific master topic, you can take it and contact us at mairo.leier_at_taltech.ee, uljana.reinsalu_at_taltech.ee or karl.janson_at_taltech.ee for further discussion.
Topics for BSc and MSc students
Integration of LLM as a UAV co-pilot
(Added 06.02.2026)
Status: Active
Description:
TalTech EAI Lab is using custom drone with additional cameras and Jetson Orin Nano that could be used for different computationally heavy tasks. This project explores the use of a Large Language Model (LLM) as a UAV co-pilot in a simulation-based research setting, with optional testing on UAV hardware. The goal is to study how LLMs can assist a human pilot at a high-level decision and interaction layer.
Explanation:
- Interpret high-level goals
- Suggest actions or mission steps
- Respond to natural-language pilot commands
- Provide explanations of system behaviour
Expected Outcomes:
- A simulation-based prototype of an LLM-powered UAV co-pilot
- Demonstration of natural language interaction with a simulated UAV
- Initial evaluation of LLM reasoning for mission-level decision support
- Optional hardware-in-the-loop tests to validate system integration
Signal propagation analysis for drone communication reliability inspection
(Added 06.02.2026)
Status: Active
Description:
When a drone operates at increasing distances from the operator, ground station, or docking system, the communication link may degrade or be lost due to terrain elevation, buildings, vegetation, background RF noise, or other environmental factors. Operators are typically notified only after signal strength becomes weak or communication is lost. This project aims to analyze and predict potential communication issues in advance by evaluating signal propagation along a known or planned drone trajectory.
Explanation:
The project focuses on RF signal propagation analysis using geospatial data, such as terrain elevation models and construction data provided by Maa-amet. By combining planned flight trajectories with GIS data, the student will estimate line-of-sight (LOS) and signal strength degradation over distance and terrain.
Existing libraries and methods for RF propagation and LOS analysis may be used, such as viewshed-based approaches. An RF signal viewshed determines which areas can receive a signal from a transmitter based on terrain and obstacles.
Example implementation of LOS analysis:
https://3d.gisvis.ee/
The student will be guided in selecting test routes, defining communication scenarios, and identifying suitable GIS datasets for analysis.
Expected Outcomes:
- Implementation of an RF signal propagation or LOS analysis algorithm similar to the referenced prototype
- Visual representation of signal strength or predicted signal loss over a predefined flight area or trajectory
- Integration of terrain and environmental data into the analysis
- A working prototype implemented as a web-based service or a Python-based application
- Demonstration of how predicted communication reliability changes along different drone routes
Machine-Laerning Based Predictive Maintenance and Health Monitoringof UAV
(Added 06.02.2026)
Status: Active
Description:
This project investigates unstable and unpredictable drone behavior caused by excessive vibration and shaking during flight. The focus is on situations where the Inertial Measurement Unit (IMU) reaches its operational limits (e.g. saturation, clipping, or degraded signal quality), leading to incorrect state estimation and unsafe control actions such as sudden altitude increase or loss of stability.
Explanation:
The student will collect and analyze real IMU data using a custom drone provided by the lab, supported by existing software tools for data logging and flight operation. Additional datasets may be generated through simulation. The student must define realistic vibration-inducing scenarios (e.g. mechanical imbalance, aggressive maneuvers, environmental disturbances) and design an experiment plan to reproduce them safely.
The work includes analyzing how vibration affects IMU measurements, how these errors propagate through estimation and control algorithms, and under which conditions the drone behavior becomes unstable. Depending on the number of students involved, the resulting models or detection methods will be implemented and validated either in a Gazebo-based simulation environment or directly on the drone.
- Structural or vibration-related indicators of airframe health
- Motor and propeller condition inferred from power, RPM, and vibration signals
- Sensor diagnostics, including drift, noise increase, or intermittent failures
- Pre-flight readiness classification (e.g., safe to fly, warning, maintenance required)
Expected Outcomes:
- Definition and execution of experimental scenarios that induce vibration-related sensor degradation
- Collection and evaluation of real IMU flight data from the custom drone
- Analysis of IMU limit violations and their impact on flight stability and control behavior
- Development of a model or method to detect or characterize unsafe vibration conditions
- Implementation and validation in simulation (Gazebo) and/or on the real drone
- Demonstration of results through experiments, simulations, or controlled flight tests
PhD research topics
Swarm based formation and navigation
Coordinated multi-agent localization, formation control, and navigation enabling scalable and resilient swarm operations. Real-time environment perception, object detection, tracking, and risk assessment to support safe and intelligent autonomous decision making.