NOD-TAMP: Generalizable Long-Horizon Planning with Neural Object Descriptors

TL;DR

NOD-TAMP is a framework combining neural object descriptors with task and motion planning to enable generalizable, long-horizon manipulation in complex, diverse environments, demonstrated on real-world tasks.

Contribution

The paper introduces NOD-TAMP, a novel integration of neural object descriptors with TAMP for scalable, generalizable long-horizon manipulation planning.

Findings

Outperforms prior NOD-based methods on tabletop tasks

Successfully generalizes to diverse, contact-rich manipulation tasks

Effective in real-world tool-use and high-precision insertion scenarios

Abstract

Solving complex manipulation tasks in household and factory settings remains challenging due to long-horizon reasoning, fine-grained interactions, and broad object and scene diversity. Learning skills from demonstrations can be an effective strategy, but such methods often have limited generalizability beyond training data and struggle to solve long-horizon tasks. To overcome this, we propose to synergistically combine two paradigms: Neural Object Descriptors (NODs) that produce generalizable object-centric features and Task and Motion Planning (TAMP) frameworks that chain short-horizon skills to solve multi-step tasks. We introduce NOD-TAMP, a TAMP-based framework that extracts short manipulation trajectories from a handful of human demonstrations, adapts these trajectories using NOD features, and composes them to solve broad long-horizon, contact-rich tasks. NOD-TAMP solves existing manipulation benchmarks with a handful of demonstrations and significantly outperforms prior NOD-based approaches on new tabletop manipulation tasks that require diverse generalization. Finally, we deploy NOD-TAMP on a number of real-world tasks, including tool-use and high-precision insertion. For more details, please visit https://nodtamp.github.io/.