Efficient Edge AI & Intelligent Perception

We develop resource-efficient machine learning and perception algorithms that bring advanced AI capabilities to embedded and robotic platforms. Our research emphasizes hardware-aware neural architecture search, energy-efficient computer vision, model optimization, and real-time edge intelligence for autonomous systems with limited computing resources.

Hardware-aware neural architecture search for object detection on microcontrollers
Once-for-all training that trades accuracy and robustness at run time
Efficient waste detection for multi-robot clean-up
Embedded language-model guidance and energy-aware learning on mobile devices

Representative publications

Lab members are in bold. ★ PhD student, ♦ Master's student, † equal contribution. The full list is on the Publications page.

2026

WACV-WS

TrashDet: Iterative Neural Architecture Search for Efficient Waste Detection

Tony Tran^♦ and Bin Hu

In WASTEVISION: International Workshop on Smart Waste Monitoring at IEEE/CVF Winter Conf. on Applications of Computer Vision (WACV-WS), 2026

Graph Bib PDF Project

@inproceedings{tran2026trashdet,
  title = {TrashDet: Iterative Neural Architecture Search for Efficient Waste Detection},
  author = {Tran♦, Tony and Hu, Bin},
  booktitle = {WASTEVISION: International Workshop on Smart Waste Monitoring at IEEE/CVF Winter Conf. on Applications of Computer Vision},
  year = {2026},
  address = {Tucson, Arizona, USA},
}

IEEE TC

ELASTIC: Efficient Once For All Iterative Search for Object Detection on Microcontrollers

Tony Tran^♦, Qin Lin, and Bin Hu

IEEE Transactions on Computers (IEEE TC), 2026

Graph Bib PDF Code Project

@article{tran2026elastic,
  title = {ELASTIC: Efficient Once For All Iterative Search for Object Detection on Microcontrollers},
  author = {Tran♦, Tony and Lin, Qin and Hu, Bin},
  journal = {IEEE Transactions on Computers},
  year = {2026},
  pages = {2281--2289},
  keywords = {Optimization;Iterative methods;Computer architecture;Object detection;Neck;Search problems;Detectors;Tiny machine learning;Microcontrollers;Costs;Neural Architecture Search;TinyML;Object Detection;Population Passthrough;Iterative Evolutionary Architecture Search},
  doi = {10.1109/TC.2026.3678184},
  volume = {75}
}

WACV

SOLAR: Switchable Output Layer for Accuracy and Robustness in Once-for-All Training

Shaharyar Ahmed Khan Tareen^★, Lei Fan, Xiaojing Yuan, Qin Lin, and Bin Hu

In Proc. IEEE/CVF Winter Conf. on Applications of Computer Vision (Round 2 New Submission Acceptance Rate: 26.88%) (WACV), 2026

Graph Bib PDF Code Project

@inproceedings{tareen2026solar,
  title = {SOLAR: Switchable Output Layer for Accuracy and Robustness in Once-for-All Training},
  author = {Tareen★, Shaharyar Ahmed Khan and Fan, Lei and Yuan, Xiaojing and Lin, Qin and Hu, Bin},
  booktitle = {Proc. IEEE/CVF Winter Conf. on Applications of Computer Vision (Round 2 New Submission Acceptance Rate: 26.88%)},
  year = {2026},
  address = {Tucson, Arizona, USA},
}

2025

ICRA-LBS

Toward Embedded LLM-Guided Navigation and Object Detection for Aerial Robots

Richie R Suganda^★ and Bin Hu

In 2025 IEEE International Conference on Robotics and Automation Late Breaking Session (ICRA-LBS), 2025

Graph Bib Poster Demo

@inproceedings{llm2025aerial,
  title = {Toward Embedded LLM-Guided Navigation and Object Detection for Aerial Robots},
  author = {Suganda★, Richie R and Hu, Bin},
  booktitle = {2025 IEEE International Conference on Robotics and Automation Late Breaking Session},
  year = {2025},
  address = {Atlanta, USA},
}

2022

ICDCS

Energy Minimization for Federated Asynchronous Learning on Battery-Powered Mobile Devices via Application Co-running

Cong Wang, Bin Hu, and Hongyi Wu

In 2022 IEEE International Conference on Distributed Computing Systems (Acceptance Rate: 19.9%) (ICDCS), 2022

Graph Bib PDF

@inproceedings{wang2022energy,
  title = {Energy Minimization for Federated Asynchronous Learning on Battery-Powered Mobile Devices via Application Co-running},
  author = {Wang, Cong and Hu, Bin and Wu, Hongyi},
  booktitle = {2022 IEEE International Conference on Distributed Computing Systems (Acceptance Rate: 19.9%)},
  year = {2022},
  pages = {939--949},
  address = {Bologna, Italy}
}

Funded projects

NVIDIA Academic Grant Program Heterogeneous Multi-Robot Waste Detection with Conformal Runtime Monitoring 2026 PI