Jianwei Zhu

Jiangsu University of Science and Technology

Papers

Total Citations

H-Index

About

Jianwei Zhu is a rising star in robotic manipulation, whose work sits at the intersection of computer vision, deep learning, and human-robot interaction. His primary research focuses on robotic grasp detection and task-oriented grasping (TOG), aiming to give robots the dexterity and intelligence to handle unknown objects and follow complex human commands. Zhu’s most notable contribution is the **HTC-Grasp** architecture, a pioneering hybrid Transformer-CNN model that significantly improves the accuracy of detecting viable grasp poses for unfamiliar objects. By integrating a hierarchical transformer with external attention, his work addresses a fundamental challenge in robotics, earning 21 citations for the 2023 paper. Building on this, his latest 2025 work, **VLA-Grasp**, pushes the frontier further by introducing a vision-language-action model with cross-modality fusion. This allows robots to not only see an object but also understand a specific linguistic task (e.g., “grip the handle”) to predict the correct grasp pose. Though still nascent with 3 citations, this work signals a major step toward more intuitive, language-guided robotic assistants. Zhu’s research is critical for advancing autonomous systems in manufacturing, logistics, and service robotics.