Publications

Research across robotics, learning, and systems.

Publications and preprints in safe robot autonomy, bipedal navigation, efficient multimodal inference, and learning-assisted optimization. An asterisk denotes equal contribution.

Google Scholar Semantic Scholar OpenReview

2026

Accepted · Selected publication

RedVTP: Training-Free Acceleration of Diffusion Vision-Language Models Inference via Masked Token-Guided Visual Token Pruning

Jingqi Xu*, Jingxi Lu*, Chenghao Li*, Sreetama Sarkar, Souvik Kundu, Peter A. Beerel

CVPR 2026 Findings

A response-driven visual-token pruning method that accelerates diffusion vision-language model inference while preserving benchmark accuracy.

Paper Code

Published · Selected publication

STATE-NAV: Stability-Aware Traversability Estimation for Bipedal Navigation on Rough Terrain

Ziwon Yoon, Lawrence Y. Zhu, Jingxi Lu, Lu Gan, Ye Zhao

IEEE Robotics and Automation Letters, vol. 11, no. 2, pp. 2338–2345

A stability-aware navigation framework that learns bipedal instability and integrates predicted safe velocity into hierarchical rough-terrain planning.

Project Paper DOI Video

Published · Selected publication

Hybrid Learning for Cold-Start-Aware Microservice Scheduling in Dynamic Edge Environments

Jingxi Lu, Wenhao Li, Jianxiong Guo, Xingjian Ding, Zhiqing Tang, Tian Wang, Weijia Jia

IEEE Transactions on Mobile Computing, vol. 25, no. 5, pp. 7338–7353

Behavior cloning and Soft Actor-Critic address cold-start learning, interdependent scheduling decisions, and dynamic edge resources.

Paper DOI Code

Accepted

Graph Neural Networks with Diversity-Aware Neighbor Selection and Dynamic Multi-Scale Fusion for Multivariate Time Series Forecasting

Jingqi Xu, Guibin Chen, Jingxi Lu, Yuzhang Lin

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Diversity-aware neighbor selection reduces redundant graph information while dynamic fusion combines temporal features across scales.

Paper

2025

Preprint

HIVTP: A Training-Free Method to Improve VLMs Efficiency via Hierarchical Visual Token Pruning Using Middle-Layer-Based Importance Score

Jingqi Xu*, Jingxi Lu*, Chenghao Li, Sreetama Sarkar, Peter A. Beerel

arXiv preprint

Middle-layer attention and hierarchical global-local token selection retain fine-grained visual information while reducing VLM inference cost.

Paper Code

Published

A Reinforcement Learning Approach for Initialization of Column Generation with Application to Aircraft Recovery Problem

Jingxi Lu, Xiongwen Qian

10th International Conference on Cyber Security and Information Engineering (ICCSIE)

PPO, graph attention, and pointer networks generate initial columns that accelerate aircraft-recovery optimization.

DOI

2024

Published

Container Scheduling with Dynamic Computing Resource for Microservice Deployment in Edge Computing

Jingxi Lu, Wenhao Li, Jianxiong Guo, Xingjian Ding, Zhiqing Tang, Tian Wang

20th International Conference on Mobility, Sensing and Networking (MSN), pp. 236–243

A GRU-enhanced Soft Actor-Critic scheduler for dynamic edge resources and interdependent microservice decisions.

DOI