Pavlos Tosidis

About

Pavlos Tosidis is a researcher at the forefront of integrating deep learning with active robotic perception, focusing on how robots can intelligently interact with their environments to improve sensing and decision-making. His work centers on three key areas: active vision, reinforcement learning for robotics, and realistic data generation. Tosidis’s most notable contribution is his pioneering use of deep reinforcement learning to develop active vision control policies for face recognition, enabling robots to dynamically adjust their viewpoints for optimal perception—a paradigm shift from static dataset training. His research on differential-drive robot navigation using low-cost sensors, employing action masking in reinforcement learning, demonstrates a practical approach to affordable robotics. Additionally, his framework for realistic data generation via deep learning-based human digitization addresses the critical need for synthetic training data. With his most-cited paper accumulating 9 citations since 2022, Tosidis’s work is gaining traction in the robotics community. His contributions to the OpenDR project further highlight his commitment to open-source tools, making advanced deep learning for robotics accessible to researchers and practitioners alike.

Research Focus

Key Achievements

Top Papers

1. 1
   Active Vision Control Policies for Face Recognition using Deep Reinforcement Learning

   9 citations · 2022
2. 2
   Efficient Realistic Data Generation Framework Leveraging Deep Learning-Based Human Digitization

   3 citations · 2021
3. 3
   Deep Learning for Active Robotic Perception

   2 citations · 2023
4. 4
   Deep learning for robotics examples using OpenDR

   2 citations · 2022
5. 5
   Deep Reinforcement Learning With Action Masking for Differential-Drive Robot Navigation Using Low-Cost Sensors

   2 citations · 2023

Key Collaborators

Contact & Links