PhD Thesis: Enhancing Robot Social Navigation

Thesis Overview

Completed PhD thesis integrating all three proposed models into a comprehensive framework for social robot navigation:

• Predictive World Models (2StepAhead, MASPM)
• Cosine-Gated LSTM architecture
• Adaptive Prediction Horizons (entropy-driven)

Environments Used

• SocNavGym (team contribution)
• FallingBallEnv (created)
• LiteSocNavGym (created)
• LunarLander-v2 (benchmark)

Results

• Successfully defended PhD thesis
• 2 conference papers published (UKCI 2023, IEEE PRML 2024)
• Superior performance in all tested environments
• Comprehensive research framework established

Links:

• Thesis PDF (ZIP)
• Thesis Document (PDF)
• Viva Presentation