|
/twan ŋwɪn/ I am currently an AI Research Resident at FSoft AI Center, working closely with Prof. Anh Nguyen. My research lies in the intersection of Robotics, Multimodal Learning, and Generative Modeling. Previously, I was an Undergraduate Research Assistant at AISIA Lab. I graduated with a bachelor's degree in Computer Science at Ho Chi Minh City University of Science. |
|
|
In the long term, I envision and yearn for a world where robots assist us in every aspect of daily life. As a football lover, I am especially excited about a future where robots can not only dexterously and effectively play sports like football with us, but also coach us to improve our skills. A recent research by DeepMind on table tennis has fueled my excitement even more. In the short term, I believe that building world models is a critical step in vastly enriching the data needed for robot training, with generative models playing a key role. The recent work by 1X has given me so much renewed hope for this future. |
|
|
|
* indicates equal contribution |
|
Toan Nguyen, Minh Nhat Vu, Baoru Huang, An Vuong, Quan Vuong, Dung Nguyen, Ngan Le, Thieu Vo, Anh Nguyen European Conference on Computer Vision (ECCV), 2024, Oral [arXiv] [Project] [Code] We introduce a novel diffusion model incorporating the new concept of negative prompt guidance learning to tackle the task of 6-DoF grasp detection in cluttered point clouds. |
|
|
An Vuong, Toan Nguyen, Minh Nhat Vu, Baoru Huang, Dung Nguyen, Binh Huynh, Thieu Vo, Anh Nguyen IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024 [arXiv] [Project] [Code] We introduce HabiCrowd, a new dataset and benchmark for crowd-aware visual navigation that surpasses other benchmarks in terms of human diversity and computational utilization. |
|
Toan Nguyen, Minh Nhat Vu, Baoru Huang, Tuan Vo, Vy Truong, Ngan Le, Thieu Vo, Bac Le, Anh Nguyen IEEE International Conference on Robotics and Automation (ICRA), 2024 [arXiv] [Project] [Code] We address the task of language-driven affordance-pose detection in 3D point clouds. Our method simultaneously detect open-vocabulary affordances and generate affordance-specific 6-DoF poses. |
|
Tuan Vo, Minh Nhat Vu, Baoru Huang, Toan Nguyen, Ngan Le, Thieu Vo, Anh Nguyen IEEE International Conference on Robotics and Automation (ICRA), 2024 [arXiv] [Code] We introduce a new open-vocabulary affordance detection method using knowledge distillation and text-point correlation. |
|
An Vuong, Minh Nhat Vu, Toan Nguyen, Baoru Huang, Dung Nguyen, Thieu Vo, Anh Nguyen Conference on Neural Information Processing Systems (NeurIPS), 2023 [arXiv] [Project] [Code] We introduce Language-Driven Scene Synthesis task, which involves the leverage of human-input text prompts to generate physically plausible and semantically reasonable objects. |
|
|
Toan Nguyen, Minh Nhat Vu, An Vuong, Dung Nguyen, Thieu Vo, Ngan Le, Anh Nguyen IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, Best Overall & Best Student Paper Awards Finalist [arXiv] [Project] [Code] Our method detects potentially unlimited textual affordance labels. |
|
Toan Nguyen, Huy Nguyen, Huy Ung, Hieu Ung, Binh Nguyen Under Review [arXiv] [Code] DeWi obtains state-of-the-art performances on insect pest classification benchmarks by combining image data augmentation and representation learning. |
|
|
|
|