Dongwon Park

Ulsan National Institute of Science and Technology

Papers

Total Citations

H-Index

About

Dongwon Park is a leading researcher in robotic perception and manipulation, with a primary focus on deep learning-based grasp detection and object reasoning. His work addresses fundamental challenges in enabling robots to interact with novel objects, particularly through innovative neural network architectures. Park’s seminal 2018 paper, "Classification based Grasp Detection using Spatial Transformer Network" (32 citations), introduced a novel approach that leverages spatial transformer networks to improve classification-based grasp detection, significantly enhancing a robot’s ability to handle unfamiliar objects. Building on this, his 2020 work, "A Single Multi-Task Deep Neural Network with Post-Processing for Object Detection with Reasoning and Robotic Grasp Detection" (11 citations), pioneered a unified framework that integrates object detection, relational reasoning, and grasp planning within a single deep neural network. This multi-task architecture, combined with intelligent post-processing, enables robots to not only detect objects but also understand their spatial relationships for more context-aware grasping. Park’s contributions are pivotal for advancing autonomous robotics, and his research continues to shape the development of more capable, reasoning-driven robotic systems for real-world applications.