3D Occupancy
3D occupancy prediction aims to create a detailed, three-dimensional representation of a scene's occupied and unoccupied spaces, often incorporating semantic information about objects within the scene. Current research focuses on improving the efficiency and accuracy of these predictions, exploring various model architectures including transformer-based networks, diffusion models, and those leveraging multi-sensor fusion (e.g., cameras, LiDAR, radar) to overcome limitations of individual sensor modalities. This technology is crucial for autonomous driving and robotics, enabling safer and more robust navigation by providing a comprehensive understanding of the surrounding environment.
Papers
Probabilistic Gaussian Superposition for Efficient 3D Occupancy Prediction
Yuanhui Huang, Amonnut Thammatadatrakoon, Wenzhao Zheng, Yunpeng Zhang, Dalong Du, Jiwen Lu
EmbodiedOcc: Embodied 3D Occupancy Prediction for Vision-based Online Scene Understanding
Yuqi Wu, Wenzhao Zheng, Sicheng Zuo, Yuanhui Huang, Jie Zhou, Jiwen Lu
OccRWKV: Rethinking Efficient 3D Semantic Occupancy Prediction with Linear Complexity
Junming Wang, Wei Yin, Xiaoxiao Long, Xingyu Zhang, Zebin Xing, Xiaoyang Guo, Qian Zhang
DAOcc: 3D Object Detection Assisted Multi-Sensor Fusion for 3D Occupancy Prediction
Zhen Yang, Yanpeng Dong, Heng Wang, Lichao Ma, Zijian Cui, Qi Liu, Haoran Pei