GNN Constraint Aware World Model Dataset (v3) Real robot episodes with per frame constraint graphs, SAM2 segmentation masks + 256 D feature embeddings, full 3D depth bundles, and synchronized robot states across two manipulation domains. Both domains share the v3 on disk layout (same JSON/NPZ schemas, same delta encoded frame states , same fully connected PyG expansion at load time) and now share a unified 270 D node feature format — the PyG loader reads a fixed 10 D type encoding from a YAML config so both domains produce identical node dimensionality. Project: GNN world model for constraint aware video generation Author: Texas A&M University Hardware: UR5e + Robotiq 2F 85 gripper, OAK D Pro (static side view) What's in this repo — at a glance Where Contains Use for session (Desktop) and hanoi/session hanoi Raw episodes + per frame annotations/ (masks, embeddings, depth bundles, side graph.json ) Training data for the world model config/type encoding .yaml Fixed 10 D per type encoding YAMLs Loader inputs (pick one per run) gnn world model loader.py Self contained PyG loader (one function per variant; also list all frame graphs iterator) Reading dataset → torch geometric.data.Data…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy