Talk2BEV: Language-enhanced Bird's-eye View Maps for Autonomous Driving
This addresses the need for more flexible and interpretable perception systems in autonomous driving, though it is incremental by blending existing models with BEV representations.
Talk2BEV tackles the limitation of closed-set perception in autonomous driving by integrating a large vision-language model with bird's-eye view maps, enabling a single system to handle diverse tasks like visual reasoning and intent prediction, achieving strong performance on a new benchmark with over 20,000 questions.
Talk2BEV is a large vision-language model (LVLM) interface for bird's-eye view (BEV) maps in autonomous driving contexts. While existing perception systems for autonomous driving scenarios have largely focused on a pre-defined (closed) set of object categories and driving scenarios, Talk2BEV blends recent advances in general-purpose language and vision models with BEV-structured map representations, eliminating the need for task-specific models. This enables a single system to cater to a variety of autonomous driving tasks encompassing visual and spatial reasoning, predicting the intents of traffic actors, and decision-making based on visual cues. We extensively evaluate Talk2BEV on a large number of scene understanding tasks that rely on both the ability to interpret free-form natural language queries, and in grounding these queries to the visual context embedded into the language-enhanced BEV map. To enable further research in LVLMs for autonomous driving scenarios, we develop and release Talk2BEV-Bench, a benchmark encompassing 1000 human-annotated BEV scenarios, with more than 20,000 questions and ground-truth responses from the NuScenes dataset.