Overview | 요약
Stochastic processes and stochastic dynamical systems are widely used in many fields to model real-world systems affected by random phenomena (e.g., inherent uncertainties, external disturbances, measurement noise).
Optimal control and estimation in the presence of these random phenomena pose many unique challenges that are absent from deterministic systems.
Thus, stochastic systems remains a topic of high interest in the engineering community.
Recently, artificial intelligence (AI) has been immensely useful in the development of tools for stochastic control and estimation, including data-driven control policies and black-box modeling of highly complex dynamics.
However, two well-known issues with AI-based methods (specifically, model-free methods) are its reliance on the availability of good-quality data and lengthy training time.
While model-based methods, on the other hand, do not suffer from these issues, they often only work for systems under simple or restrictive assumptions, e.g., Gaussian white noise.
This suggests that examining the tradeoff between model-free and model-based methods gives to us a wide spectrum of methods for handling stochastic systems.
Our Goal | 목적
We develop theory and algorithms for stochastic control, filtering & estimation, and autonomous decision-making by combining traditional model-based methods with modern data-driven AI methods.
Our research spans a broad range of applications, including motion-planning, fault-tolerance, robotics, reinforcement learning, multi-agent systems, sensor networks, and embodied AI.
See our individual projects for more.
Announcement for Prospective Postdocs:
Prof. Hiroyasu Tsukamoto at UIUC and I are looking for highly-motivated and skilled postdocs to collaborate with our research team on a new government-sponsored, multi-university project at the intersection of
human-machine AI autonomy and
nonlinear stochastic control theory!
Check out the details in the LinkedIn post [HERE]!
Applications and/or inquiries may be emailed directly to soojean@kaist.ac.kr and hiroyasu@illinois.edu.
To prospective students who would like to join our lab: Thank you, first of all, for your interest.
Please consult the following instructions:
- Undergraduate Interns: Please send an email to acss-admins@kaist.ac.kr with your CV and transcript attached.
- MS/PhD Applicants: Please first apply through the EE application process [Link].
Then send an email to acss-admins@kaist.ac.kr with your CV and transcript attached.
Other inquiries may also be directed to acss-admins@kaist.ac.kr.
Any direct emails to the professor might be missed.
