AssemblyBench: Physics-Aware Assembly of Complex Industrial Objects
For researchers in robotic assembly and 3D vision, this work provides a new benchmark and method for physically plausible assembly of complex industrial objects.
AssemblyBench introduces a synthetic dataset of 2,789 industrial objects with multimodal instructions and 3D parts, and proposes AssemblyDyno, a transformer model that jointly predicts assembly order and trajectories, outperforming prior works in pose estimation and trajectory feasibility.
Assembling objects from parts requires understanding multimodal instructions, linking them to 3D components, and predicting physically plausible 6-DoF motions for each assembly step. Existing datasets focus on simplified scenarios, overlooking shape complexities and assembly trajectories in industrial assemblies. We introduce AssemblyBench, a synthetic dataset of 2,789 industrial objects with multimodal instruction manuals, corresponding 3D part models, and part assembly trajectories. We also propose a transformer-based model, AssemblyDyno, which uses the instructional manual and the 3D shape of each part to jointly predict assembly order and part assembly trajectories. AssemblyDyno outperforms prior works in both assembly pose estimation and trajectory feasibility, where the latter is evaluated by our physics-based simulations.