PIER-Flow: Physics-Informed Efficient Rectified Flow for Real-Time Mobile Robot Navigation

2026-07-11

AI Summary

This work addresses the challenge of achieving real-time, multimodal obstacle avoidance with kinematic consistency in dense, highly dynamic environments—a task where existing methods often fall short. The authors propose a physics-informed rectified flow–based policy distillation framework that encodes a model predictive control (MPC) expert policy into a continuous-time ordinary differential equation (ODE). By integrating parallel latent sampling, lightweight feasibility filtering, and an asynchronous action-chunking architecture, the method achieves millisecond-level single-step inference while preserving kinematic consistency. Experiments demonstrate a 98.85% success rate with zero collisions in simulation, at an average inference latency of just 1.29 ms—yielding a 37.2× speedup over MPC and an 800× improvement over standard diffusion models. On real-world edge hardware, the system maintains stable operation with approximately 5.3 ms latency.

Abstract

Autonomous navigation in dense and highly dynamic environments requires both physically feasible control and low-latency replanning. Optimization-based methods such as Model Predictive Control (MPC) explicitly handle robot kinematics and safety constraints, but repeated nonlinear optimization can limit real-time responsiveness. Deterministic behavior-cloning policies enable efficient inference but may fail to represent multimodal avoidance behaviors, whereas diffusion policies capture multimodality at the cost of time-consuming iterative denoising. We propose PIER-Flow (Physics-Informed Efficient Rectified Flow), a lightweight navigation policy for mobile robots. By distilling an MPC expert into a continuous-time Ordinary Differential Equation (ODE), PIER-Flow achieves single-step action generation through parallel latent sampling and lightweight feasibility selection. We introduce a physics-informed training objective to enforce kinematic consistency, paired with an asynchronous action chunking architecture for robust sim-to-real deployment. Extensive simulations demonstrate that PIER-Flow achieves a 98.85\% success rate and zero collisions, with an average inference of $\sim$1.29 ms, which accelerates planning by 37.2$\times$ compared to MPC and over 800$\times$ against standard diffusion models. Crucially, real-world deployment on a resource-constrained edge computer further achieves an approximately stable inference latency of $\sim$5.3 ms, avoiding the latency spikes and freezing events observed with planning baselines.

Problem

• real-time navigation
• mobile robot
• multimodal behavior
• low-latency planning
• physics feasibility

Innovation

• Physics-Informed Learning
• Rectified Flow
• Mobile Robot Navigation
• Real-Time Inference
• Sim-to-Real Transfer