SwarmDiff: Swarm Robotic Trajectory Planning in Cluttered Environments via Diffusion Transformer

2025-05-21

AI Summary

To address low computational efficiency, poor scalability, and insufficient safety in swarm robot trajectory planning under dense obstacle environments, this paper proposes a hierarchical generative framework. It represents the swarm’s macroscopic state via probability density functions and employs a conditional diffusion model to generate risk-aware macroscopic trajectory distributions, which guide microscopic individual trajectory optimization. Innovatively, we integrate Wasserstein optimal transport with conditional value-at-risk (CVaR) to establish an optimal transport–risk co-optimization mechanism. Furthermore, we design a Diffusion Transformer (DiT) architecture to enhance long-range dependency modeling and improve sampling efficiency. Experiments demonstrate that our method achieves a 2.3× speedup, improves trajectory validity by 18.7%, and enables real-time, safe, and robust navigation for swarms of up to 100 robots in both simulation and real-world scenarios.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Intelligent Robots: Motion and Path Planning
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Swarm robotic trajectory planning faces challenges in computational efficiency, scalability, and safety, particularly in complex, obstacle-dense environments. To address these issues, we propose SwarmDiff, a hierarchical and scalable generative framework for swarm robots. We model the swarm's macroscopic state using Probability Density Functions (PDFs) and leverage conditional diffusion models to generate risk-aware macroscopic trajectory distributions, which then guide the generation of individual robot trajectories at the microscopic level. To ensure a balance between the swarm's optimal transportation and risk awareness, we integrate Wasserstein metrics and Conditional Value at Risk (CVaR). Additionally, we introduce a Diffusion Transformer (DiT) to improve sampling efficiency and generation quality by capturing long-range dependencies. Extensive simulations and real-world experiments demonstrate that SwarmDiff outperforms existing methods in computational efficiency, trajectory validity, and scalability, making it a reliable solution for swarm robotic trajectory planning.

Problem

• Swarm robotic trajectory planning in cluttered environments
• Balancing optimal transportation and risk awareness
• Improving computational efficiency and scalability

Innovation

• Hierarchical generative framework for swarm robots
• Probability Density Functions model macroscopic state
• Diffusion Transformer enhances sampling efficiency