Panagiota Grosdouli

About

Building a research path in robust autonomous systems.

I am an Electrical and Computer Engineering student working toward a focused research profile in robotics, perception, localization, planning, and uncertainty-aware autonomy.

My portfolio is structured as a research environment rather than a project showcase. Each module is framed through a scientific question, current implementation status, limitations, and next experiments.

The central theme is robust autonomy under uncertainty: how autonomous systems can perceive, estimate, plan, and recover when their inputs are incomplete, degraded, or unreliable.

Education

Academic background

MEng Electrical & Computer Engineering

2020–2026

Democritus University of Thrace, Xanthi, Greece

Thesis direction: trajectory prediction of vulnerable road users at smart intersections, with emphasis on intelligent mobility, uncertainty, and safety-critical decision support.

Research directions

Questions guiding the work.

Robust localization

How can VIO and SLAM systems expose degradation early enough for downstream safety decisions?

Uncertainty-aware navigation

How should robots adapt plans when localization, perception, or map information becomes unreliable?

Intelligent mobility

How can trajectory prediction and interaction reasoning improve safety for vulnerable road users?

Technical profile

Tools for research software.

Robotics systems

Visual-inertial localizationSLAMsensor fusionmotion planningROS 2simulation

Learning and perception

computer visiontrajectory predictionuncertainty-aware evaluationPyTorchOpenCVdata analysis

Research engineering

PythonC/C++TypeScriptLinuxGitDockertechnical writing

Scientific practice

failure-case analysisreproducible experimentsmetric designablation planninglimitations reporting

Research principles

How the site presents work.

  • Claims should be traceable to implementation status, data, or clearly marked future work.
  • A robotics system should report uncertainty in a form that a planner or safety layer can actually use.
  • Research software is strongest when assumptions, limitations, and failure modes are visible to the reader.