Paper | Code
Our V2V-GoT-QA is a dataset for Multimodal LLM-based cooperative autonomous driving with graph-of-thoughts reasoning. V2V-GoT-QA includes 9 different types of perception, prediction, and planning QA tasks, with 110K training and 31K testing QA pairs.
Occlusion-aware Perception (Q1 - Q4): consider visible, occluding, and invisible objects. Planning-aware Prediction (Q5 - Q7): include… See the full description on the dataset page:
https://huggingface.co/datasets/eddyhkchiu/V2V-GoT-QA.
