Mixed-Density Diffuser: Efficient Planning with Non-Uniform Temporal Resolution
This work addresses performance degradation in diffusion-based planning for robotics and control tasks, offering an incremental improvement by optimizing temporal resolution.
The paper tackled the problem of diffusion planners degrading performance with excessively sparse step predictions by hypothesizing that temporal density thresholds are non-uniform across horizons, proposing the Mixed-Density Diffuser (MDD) with tunable hyperparameters for density, and achieved new state-of-the-art results on Maze2D, Franka Kitchen, and Antmaze D4RL task domains.
Recent studies demonstrate that diffusion planners benefit from sparse-step planning over single-step planning. Training models to skip steps in their trajectories helps capture long-term dependencies without additional or memory computational cost. However, predicting excessively sparse plans degrades performance. We hypothesize this temporal density threshold is non-uniform across a temporal horizon and that certain parts of a planned trajectory should be more densely planned. We propose Mixed-Density Diffuser (MDD), a diffusion planner where the densities throughout the horizon are tunable hyperparameters. We show that MDD achieves a new SOTA across the Maze2D, Franka Kitchen, and Antmaze D4RL task domains.