Compositional Diffusion-Based Continuous Constraint Solvers

TL;DR

This paper presents Diffusion-CCSP, a novel diffusion-based approach for solving continuous constraint satisfaction problems in robotics, enabling better generalization and integration into planning systems.

Contribution

Introduces a compositional diffusion model that combines energies from individual constraints to solve CCSPs, improving generalization and planning integration.

Findings

Strong generalization to new constraint combinations

Effective integration into task and motion planning

Outperforms previous hand-engineered or learned methods

Abstract

This paper introduces an approach for learning to solve continuous constraint satisfaction problems (CCSP) in robotic reasoning and planning. Previous methods primarily rely on hand-engineering or learning generators for specific constraint types and then rejecting the value assignments when other constraints are violated. By contrast, our model, the compositional diffusion continuous constraint solver (Diffusion-CCSP) derives global solutions to CCSPs by representing them as factor graphs and combining the energies of diffusion models trained to sample for individual constraint types. Diffusion-CCSP exhibits strong generalization to novel combinations of known constraints, and it can be integrated into a task and motion planner to devise long-horizon plans that include actions with both discrete and continuous parameters. Project site: https://diffusion-ccsp.github.io/