|
Dawei Shi's Homepage |
| Dr. Dawei Shi Professor & Director, Institute of Intelligent Sensing & Motion Control, School of Automation, Beijing Institute of Technology (BIT) Principle Investigator, Intelligent Control & Decision of Precision Medicine Systems, BIT Institute of Engineering Medicine Office: Room 212, Building #6 Email: daweishi at bit.edu.cn Phone: +86 10 6891 8802 Address: School of Automation, Beijing Institute of Technology, Beijing, 100081, P.R. China |
Research
Funded by the National Natural Science Foundation of China, the National Natural Science Foundation of Beijing, and various industrial collaborators, our research generally focuses on Analysis & Synthesis of Advanced Sampled-data Control Systems, with applications to Biomedical Engineering, Robotics and motion systems. Specifically, we are interested in adaptive and coordinated learning, optimization and planning in sampled-data systems, and work on the following topics:
Event-triggered sampled-data control, state estimation and machine learning
Generally, sampled-data control deals with issues caused by the hybrid nature of a digital control system, in which the controllers implemented on digital processors work in discrete time while the plants evolve in continuous time. In particular, the recently developed event-triggered/based sampling technique offers new opportunities in achieving satisfactory or even enhanced control performance with reduced sampling or transmission rates. In this background, our aim is to develop feedback controllers, state estimators and adaptive learning algorithms that can work efficiently with event-triggering information transmission/update mechanisms.
Closed-loop drug delivery systems
The development of sensors that continously monitor physiological variables and actuators that can dynamically infuse micro-dosages of drugs has enabled the research on closed-loop drug delivery systems. In this direction, we work on the development, in silico verification and clinical study of artificial pancreas systems that is used to achieve closed-loop regulation of blood glucose for subjects with type 1 diabetes. We also work on automated decision algorithms that allow more precise dosage of meal boluses. In addition, through developing wearable health monitoring systems, we design intelligent learning aglorithms to dynamically estimate the key performance metrics that reflect disease status and advanced alarm systems that predict health-threatening events.
Image-based sensing, control and decision
Image-based sensing and monitoring offer new opportunities of designing autonomous systems that can smartly achieve challenging tasks. With image-based information, we aim to develop advanced optimization-based sampled-data control algorithms that enable high-performance dynamic path planning and motion decision for wheel-legged robotic systems. Through designing deep-learning based segmentation algorithms, we also exploit spacially sampled MRI slices to design intelligent decision methods for precision diagnosis of spinal diseases.