arXiv subjects

Yukun Xie

Publications and source records attributed to Yukun Xie.

3 recordsLinked to original sources

arXiv2026

RoboFollow: Unveiling the Instruction Following Mirage in Embodied Agents

Modern embodied agents achieve impressive success rates, yet their actual instruction-following ability is far weaker than these numbers suggest. We trace this illusion to a structural property we term low scene entropy: when a visual scene admits only one valid task, language becomes redundant and a policy can score highly while barely using it. We introduce RoboFollow, a diagnostic benchmark with three principles: (1) High Scene Entropy: each training scene supports multiple kinematically distinct task branches, making vision alone insufficient and forcing reliance on language. (2) Hierarchical Diagnostic Protocol: a four-level protocol (L0--L3) progressively perturbs visual layout and semantics, probing whether equivalent instructions yield consistent behavior and distinct ones yield discriminable behavior across spatial relations, attributes, trajectory constraints, and logic. (3) Confound-Controlled Diagnosis: we simplify interaction objects, restrict actions to the trained repertoire and report stage-wise Intent and Execution scores, isolating comprehension from motor execution. Evaluation of nine VLA and WAM policies shows that strong L0 performance, where attained, does not reliably transfer to L1--L3 under our fine-tuning setup. Representative mitigations, including stronger VLM backbones, QA co-training, LangForce, and Classifier-Free Guidance, all fail to close this gap. RoboFollow exposes genuine instruction following as a critical, overlooked bottleneck. Code and dataset are available at https://github.com/AutoLab-SAI-SJTU/RoboFollow and https://huggingface.co/datasets/AutoLab-SJTU/robofollow-data.

cs.RO↗

arXiv2026

LinCa: Accelerating Diffusion Models via Learnable Decomposed Feature Caching

Diffusion models have achieved remarkable success in image and video generation, yet the high computational cost of iterative sampling remains a critical bottleneck for practical deployment. Feature caching has emerged as a promising acceleration paradigm by reusing or predicting intermediate features across timesteps. However, existing training-free methods apply uniform prediction strategies that cannot adapt to the heterogeneous feature dynamics, causing significant quality degradation under high acceleration ratios. We propose LinCa, a feature caching framework based on learnable invertible networks. LinCa decomposes cached features into sub-components with distinct continuity properties via a lightweight invertible network and applies differentiated prediction orders matched to each component. The strict invertibility guarantees lossless reconstruction back to the original feature space, forming a unified Decompose-Predict-Reconstruct pipeline. By training separate predictors for different models and timestep segments, LinCa adapts to heterogeneous feature dynamics. Experiments on FLUX, Qwen-Image, and HunyuanVideo demonstrate that LinCa, with less than 0.2% additional parameters, significantly outperforms existing methods and maintains near-lossless quality at 5-7x speedup. Code: https://github.com/QHR69/LinCa

cs.CV↗

arXiv2024

COMPILED: Deep Metric Learning for Defect Classification of Threaded Pipe Connections using Multichannel Partially Observed Functional Data

In modern manufacturing, most products are conforming. Few products are nonconforming with different defect types. The identification of defect types can help further root cause diagnosis of production lines. With the sensing technology development, process variables evolved as time changes, which can be collected in high resolution as multichannel functional data. These functional data have rich information to characterize the process and help identify the defect types. Motivated by a real example from the threaded pipe connection process, we focus on defect classification where each sample is represented as partially observed multichannel functional data. However, the available samples for each defect type are limited and imbalanced. The functional data is partially observed since the pre-connection process before the threaded pipe connection process is unobserved as there is no sensor installed in the production line. Therefore, the defect classification based on imbalanced, multichannel, and partially observed functional data is very important but challenging. To deal with these challenges, we propose an innovative classification approach named as COMPILED based on deep metric learning. The framework leverages the power of deep metric learning to train on imbalanced datasets. A novel neural network structure is proposed to handle multichannel partially observed functional data. The results from a real-world case study demonstrate the superior accuracy of our framework when compared to existing benchmarks.

cs.LG↗