DRIFT: Diffusion-based Rule-Inferred For Trajectories

2026-03-01

AI Summary

This work addresses the challenge of generating mobile robot trajectories in unstructured environments, where achieving both global smoothness and terminal precision remains difficult. The authors propose a conditional diffusion-based trajectory generation framework that uniquely integrates graph neural network–driven structured scene perception with a graph-conditioned, time-aware GRU (GTGRU). By embedding relational inductive biases and temporal attention mechanisms into the diffusion model, the approach jointly optimizes trajectory smoothness and local accuracy. Experimental results demonstrate that the method achieves a final displacement error (FDE) of 0.041 meters and a low jerk value of 27.19 in trajectory imitation tasks, producing reference trajectories that exhibit both high fidelity and execution feasibility.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Trajectory generation for mobile robots in unstructured environments faces a critical dilemma: balancing kinematic smoothness for safe execution with terminal precision for fine-grained tasks. Existing generative planners often struggle with this trade-off, yielding either smooth but imprecise paths or geometrically accurate but erratic motions. To address the aforementioned shortcomings, this article proposes DRIFT (Diffusion-based Rule-Inferred for Trajectories), a conditional diffusion framework designed to generate high-fidelity reference trajectories by integrating two complementary inductive biases. First, a Relational Inductive Bias, realized via a GNN-based Structured Scene Perception (SSP) module, encodes global topological constraints to ensure holistic smoothness. Second, a Temporal Attention Bias, implemented through a novel Graph-Conditioned Time-Aware GRU (GTGRU), dynamically attends to sparse obstacles and targets for precise local maneuvering. In the end, quantitative results demonstrate that DRIFT reconciles these conflicting objectives, achieving centimeter-level imitation fidelity (0.041m FDE) and competitive smoothness (27.19 Jerk). This balance yields highly executable reference plans for downstream control.

Problem

• trajectory generation
• kinematic smoothness
• terminal precision
• mobile robots
• unstructured environments

Innovation

• conditional diffusion
• relational inductive bias
• temporal attention
• graph neural network
• trajectory generation