Instant Neural Graphics Primitives with a Multiresolution Hash Encoding
很重要的编码优化论文，MHE的概念：
ConceptGraphs= Open-Vocabulary 3D Scene Graphs for Perception and Planning
通过LLM来判断位置关系，以此构建scene graph
还是只能判断object-level空间关系，做不了part-level manipulation
SayPlan= Grounding Large Language Models using 3D Scene Graphs for Scalable Robot Task Planning
主要的思想都在上面这个伪代码里，通过只展开部分场景图（严格层级结构），来控制输入llm的场景图大小。
A scalable approach to ground LLM-based task planners across environments spanning multiple rooms and floors
Scene Graph 通过networkx (python package)表示
Three key innovations:
- 通过
Collapsed 3DSG来在少数根节点上寻找task-relevant子图（后续通过展开子图进行进一步的搜寻），提高了scalability（避免过于复杂的整体场景图超过LLM的token限制） - 环境中任务计划的horizon会随着给定任务的复杂性而增长，LLM会倾向于产生幻觉或者不可行的动作序列。所以通过成熟的path planner such as Dijkstra来连接high-level nodes。
- An iterative replanning pipeline in order to correct for any unexecutable actions
- Missing to open the fridge before putting something into it
因此，避免由于环境本身的物理限制和谓词的矛盾，幻觉或不一致而导致的计划失败。
- Missing to open the fridge before putting something into it
Insight
- 每一次场景节点的展开与否，该节点是否是任务关注的节点都是由LLM决定的，这一点和我的想法一致。等于是将LLM作为一个检查器一层层遍历查找任务的兴趣点。
Scene Graph Simulator作为任务是否可行的验证器。
Clio= Real-time Task-Driven Open-Set 3D Scene Graphs
贡献：
- The first contribution of this paper is to propose a task-driven 3D scene understanding problem, where the robot is given a list of tasks in natural language, and has to select the granularity and the subset of objects and scene structure to retain in its map that is sufficient to complete the tasks.
- The second contribution is an algorithm for task-driven 3D scene understanding based on an Agglomerative IB approach, that is able to cluster 3D primitives in the environment into taskrelevant objects and regions
- 基于以上，实现了一个实时的pipeline
提出了针对不同任务需要不同粒度的语义信息，本文是通过结合SAM和[[CLIP多模态预训练模型]]实现，但是忽略了物体之间的谓语关系或者父子关系。本质还是智能做导航，拾取，放下，导航的基本操作。
Representation Learning for Scene Graph Completion via Jointly Structural and Visual Embedding
 Representation Learning for Scene Graph Completion via Jointly Structural and Visual Embedding/Pasted_image_20250318162533.png)
The architecture of RLSV is a three-layered hierarchical projection that projects a visual triple onto the attribute space, the relation space, and the visual space in order.
 Visual Translation Embedding Network for Visual Relation Detection/Pasted_image_20250318155431.png)
 Visual Translation Embedding Network for Visual Relation Detection/Pasted_image_20250318155444.png)
Factorizable Net= An Efficient Subgraph-based Framework for Scene Graph Generation
我的想法是将场景进行panoptic segmentation 之后再在每个物体上进行hierarchical part relation detection，异曲同工。
RelTR= Relation Transformer for Scene Graph Generation
RelTR是自下而上的方法, 使用基于Transformer的 object detector（例如DETR）生成对象候选者，然后使用relation transformer来预测object pairs之间的关系。它还设计了一种基于积分的关系表示方法，该方法将关系编码为二维矢量场。
