Enhancing Action Detection for Robot Curiosity: Integrating Online and Offline Learning

Aufgabenstellung / Topic

This thesis builds upon a previous master's thesis on online action learning and aims to improve the approach by integrating both online and offline action detection. The goal is to refine the system to better detect when a robot should exhibit curiosity based on human activity. Improvements will focus on optimizing classification accuracy, reducing latency, and enhancing adaptability. Additionally, a user study will be conducted to evaluate the system's effectiveness in real-world human-robot interaction scenarios.

You will:

• Review literature on action recognition and curiosity-driven learning
• Analyze and improve the existing online action learning approach
• Develop and integrate a combined online-offline detection model
• Implement improvements and optimize system performance
• Conduct a user study to evaluate the system
• Summarize findings in a thesis
• (Optional) Co-author a research paper

You need:

• Experience with Machine Learning and Computer Vision
• Strong programming skills in Python
• Familiarity with ROS (Robot Operating System)
• Knowledge of study design and data analysis
• Strong statistical evaluation skills
• Strong English communication skills

References

• Automatic Class Discovery and One-Shot Interactions for Acoustic Activity Recognition