Research Areas

Spatiotemporal Tubes & Approximation-free Control

Key Works:

Control Barrier Functions

There are many complex cyber-physical systems where the precise model description can not be derived analytically and therefore one cannot use model-based techniques in these situations. In such cases, due to advances in sensor and processing technologies, one can take advantage of data-driven approaches from machine learning to address these problems. Here, we aim to provide probabilistic guarantees that the system trajectories of unknown dynamical systems satisfy complex specifications. Read more

Control of Multi-Agent Systems

Multi-Robot Systems (MRS) allows the use of distributed resources (such as communication bandwidth, multiple small-robots, power, etc.) to achieve some global tasks (consensus, formation control, connectivity maintenance, and collision avoidance). Although the distributed control of multi-agent systems has been well-studied for several decades, recent technological advancements pose new challenges. At the FOCAS lab, we identify these new challenges and provide efficient solutions to these futuristic applications of multi-robot systems.

Abstraction-based Control

Though significant progress has been made in the field of control theory, its practical application is sometimes hindered by the failure to account for physical constraints. Constraints dominate the majority of real-time control challenges. In daily life, there are lots of uncertain dynamic systems where constraints appear in different forms, such as performance, saturation, physical stoppages, and safety specifications. For such systems, while designing the controller in real time, the constraints are ineludible. Read more

Control of Cyber-Physical Systems

Modern physical processes need to interact tightly with computational/digital components such as navigation systems and computational devices. These systems are usually described by the term cyber-physical system (CPS). Models of CPS are by nature complex and hybrid in the sense that they combine discrete transition systems for computational parts with continuous differential equations that represent physical processes. This makes the design of the controller for CPS very challenging. Read more

Safe Robot Manipulation

Though significant progress has been made in the field of control theory, its practical application is sometimes hindered by the failure to account for physical constraints. Constraints dominate the majority of real-time control challenges. In daily life, there are lots of uncertain dynamic systems where constraints appear in different forms, such as performance, saturation, physical stoppages, and safety specifications. For such systems, while designing the controller in real time, the constraints are ineludible. Read more

Resilience, Trust, and Awareness in Autonomous Systems

Modern physical processes need to interact tightly with computational/digital components such as navigation systems and computational devices. These systems are usually described by the term cyber-physical system (CPS). Models of CPS are by nature complex and hybrid in the sense that they combine discrete transition systems for computational parts with continuous differential equations that represent physical processes. This makes the design of the controller for CPS very challenging. Read more

Math-to-Sim-to-Real Transfer

Though significant progress has been made in the field of control theory, its practical application is sometimes hindered by the failure to account for physical constraints. Constraints dominate the majority of real-time control challenges. In daily life, there are lots of uncertain dynamic systems where constraints appear in different forms, such as performance, saturation, physical stoppages, and safety specifications. For such systems, while designing the controller in real time, the constraints are ineludible. Read more