Constraint-Aware Diffusion Models for Trajectory Optimization
This work addresses constraint violations in trajectory optimization for robotics and autonomous systems, representing an incremental improvement over existing diffusion models.
The paper tackles the problem of constraint violations in diffusion models for trajectory optimization by introducing a constraint-aware diffusion model with a novel hybrid loss function. The model outperforms traditional diffusion models in minimizing constraint violations while generating samples close to locally optimal solutions, as demonstrated on tabletop manipulation and two-car reach-avoid problems.
The diffusion model has shown success in generating high-quality and diverse solutions to trajectory optimization problems. However, diffusion models with neural networks inevitably make prediction errors, which leads to constraint violations such as unmet goals or collisions. This paper presents a novel constraint-aware diffusion model for trajectory optimization. We introduce a novel hybrid loss function for training that minimizes the constraint violation of diffusion samples compared to the groundtruth while recovering the original data distribution. Our model is demonstrated on tabletop manipulation and two-car reach-avoid problems, outperforming traditional diffusion models in minimizing constraint violations while generating samples close to locally optimal solutions.