Sampling Constrained Trajectories Using Composable Diffusion Models
Keywords: Trajectory Optimization, Amortized Approximate Inference, Generative Models
TL;DR: We use diffusion models to learn a distribution over constraint-satisfying trajectories, which we use to improve trajectory optimiziation. We use composable diffusion models to generalize the model to compositions of multiple constraints.
Abstract: Trajectory optimization and optimal control are powerful tools for synthesizing complex robot behavior using appropriate cost functions and constraints. However, methods for solving the optimization problem are often prone to local minima and sensitive to initialization. Casting trajectory optimization as an inference problem can alleviate some of these issues by generating distributions over solutions. However, the resulting inference problem can be costly. In this work, we present an approach for using diffusion models to learn a distribution over constraint-satisfying low-cost trajectories. This learned distribution is then used as the initialization for an inference-based trajectory optimization algorithm. We exploit the composability of diffusion models to generalize the learned generative model to out-of-distribution constraints which consist of the composition of multiple in-distribution constraints. We demonstrate the benefit of our approach by showing improvement over baselines on a constrained 12DoF Quadrotor task and a 7DoF robot manipulator
Submission Number: 5