Adaptive Linear Path Model-Based Diffusion

TL;DR

This paper introduces a new diffusion model framework called LP-MBD that simplifies parameter tuning and enhances adaptability, with an adaptive extension ALP-MBD that uses reinforcement learning for improved robustness and real-time performance in robotic tasks.

Contribution

The paper proposes LP-MBD, a geometrically interpretable diffusion model with reduced tuning complexity, and ALP-MBD, an adaptive version that leverages reinforcement learning for dynamic parameter adjustment.

Findings

LP-MBD simplifies diffusion scheduling while maintaining performance.

ALP-MBD improves robustness and adaptability in robotic tasks.

Both models enhance real-time efficiency in trajectory tracking.

Abstract

The interest in combining model-based control approaches with diffusion models has been growing. Although we have seen many impressive robotic control results in difficult tasks, the performance of diffusion models is highly sensitive to the choice of scheduling parameters, making parameter tuning one of the most critical challenges. We introduce Linear Path Model-Based Diffusion (LP-MBD), which replaces the variance-preserving schedule with a flow-matching-inspired linear probability path. This yields a geometrically interpretable and decoupled parameterization that reduces tuning complexity and provides a stable foundation for adaptation. Building on this, we propose Adaptive LP-MBD (ALP-MBD), which leverages reinforcement learning to adjust diffusion steps and noise levels according to task complexity and environmental conditions. Across numerical studies, Brax benchmarks, and mobile-robot trajectory tracking, LP-MBD simplifies scheduling while maintaining strong performance, and ALP-MBD further improves robustness, adaptability, and real-time efficiency.