【論文要約:自動運転関連】Multimodal Foundational Models for Unsupervised 3D General Obstacle Detection