DDAT: Diffusion Policies Enforcing Dynamically Admissible Robot Trajectories

2025-02-20

AI Summary

To address the problem that diffusion-model-generated trajectories often violate robotic dynamical constraints, this paper proposes Dynamically Admissible Trajectory Diffusion (DATD). DATD explicitly enforces dynamical feasibility during each denoising step of the diffusion process by iteratively projecting latent states onto a polyhedral approximation of the reachable set manifold. Unlike existing black-box fine-tuning approaches, DATD is the first method to provably embed dynamical admissibility directly into the diffusion policy, enabling single-shot long-horizon trajectory planning. We evaluate DATD on MuJoCo multi-task benchmarks and real-world platforms—including Unitree GO1/GO2 quadrupeds and a quadrotor—demonstrating substantial improvements in dynamical feasibility and motion quality, while eliminating the need for frequent replanning inherent in conventional methods.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Intelligent Robots: Motion and Path Planning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Diffusion models excel at creating images and videos thanks to their multimodal generative capabilities. These same capabilities have made diffusion models increasingly popular in robotics research, where they are used for generating robot motion. However, the stochastic nature of diffusion models is fundamentally at odds with the precise dynamical equations describing the feasible motion of robots. Hence, generating dynamically admissible robot trajectories is a challenge for diffusion models. To alleviate this issue, we introduce DDAT: Diffusion policies for Dynamically Admissible Trajectories to generate provably admissible trajectories of black-box robotic systems using diffusion models. A sequence of states is a dynamically admissible trajectory if each state of the sequence belongs to the reachable set of its predecessor by the robot's equations of motion. To generate such trajectories, our diffusion policies project their predictions onto a dynamically admissible manifold during both training and inference to align the objective of the denoiser neural network with the dynamical admissibility constraint. The auto-regressive nature of these projections along with the black-box nature of robot dynamics render these projections immensely challenging. We thus enforce admissibility by iteratively sampling a polytopic under-approximation of the reachable set of a state onto which we project its predicted successor, before iterating this process with the projected successor. By producing accurate trajectories, this projection eliminates the need for diffusion models to continually replan, enabling one-shot long-horizon trajectory planning. We demonstrate that our framework generates higher quality dynamically admissible robot trajectories through extensive simulations on a quadcopter and various MuJoCo environments, along with real-world experiments on a Unitree GO1 and GO2.

Problem

• Generating dynamically admissible robot trajectories
• Aligning diffusion models with robot dynamics
• Enabling one-shot long-horizon trajectory planning

Innovation

• Diffusion models generate robot trajectories
• Projections ensure dynamical admissibility
• Iterative sampling improves trajectory accuracy