Summary:

Autonomous driving systems rely heavily on accurate object detection to ensure safety and efficient navigation. While current systems are trained on vast datasets containing common obstacles and objects, they often struggle with edge cases—unexpected or rare objects that haven't been frequently encountered during training. Open world object detection provides a potential solution by allowing the system to recognize and learn from new objects incrementally. This study aims to explore the feasibility and efficiency of implementing open world object detection in autonomous driving, particularly focusing on its ability to handle edge cases and integrate incremental training mechanisms.

Research questions: