D-SafeMPC: Diffusion-Driven Safe Model Predictive Control with Discrete-Time Control Barrier Functions

2026-07-12

AI Summary

Existing diffusion models struggle to satisfy safety and dynamical constraints in robotic planning and often exhibit poor convergence when integrated with model predictive control (MPC) due to low-quality initial trajectories. This work proposes a novel iterative co-design framework that directly embeds discrete-time Control Barrier Functions (CBFs) and Control Lyapunov Functions (CLFs) into the reverse denoising process of diffusion models. At each denoising step, an MPC optimization provides safety-aware guidance and warm-start initialization. To the best of our knowledge, this is the first approach to seamlessly integrate CBFs and CLFs into the diffusion generation process, substantially improving safety, task success rate, and computational efficiency. Extensive simulations and real-world experiments on a Franka manipulator demonstrate clear performance advantages over state-of-the-art baselines.

Abstract

A key limitation on the use of diffusion models in robotic planning is their inability to inherently enforce safety or dynamical constraints, which often results in physically infeasible or unsafe outputs. Hybrid approaches that employ model predictive control (MPC) to address this problem can be unstable, as poor trajectory initializations from the diffusion model prevent the MPC from converging to a safe and feasible solution. To overcome these challenges, we propose D-SafeMPC, which enhances the interaction between diffusion and control. Our method guides the reverse diffusion process with control barrier functions (CBFs) and control Lyapunov functions (CLFs) and employs an iterative-projection scheme where an MPC refines the trajectory at each denoising step. This steers sampling toward safe, goal-directed regions and provides reliable MPC warm starts. In simulations on a Franka manipulator across four scenarios (one static-obstacle and three dynamic-obstacle settings) and in a sim-to-real experiment on a physical Franka robot, D-SafeMPC improves safety, task success rates, and planning efficiency over state-of-the-art baselines. To facilitate reproducibility, our source code and experimental configurations are available in a repository at https://github.com/erdiphd/D-SafeMPC

Problem

• diffusion models
• safety constraints
• model predictive control
• control barrier functions
• robotic planning

Innovation

• Diffusion Models
• Model Predictive Control
• Control Barrier Functions
• Safe Planning
• Iterative Projection