Research

Control systems have often encountered various types of uncertainty and external disturbance. It is sometimes of primary necessity for system stability and satisfactory performance to compensate for those unexpected factors, for which CDSL@UoS is developing robust control theory.

The disturbance observer (DOB) plays a key role in this direction of our research. Our goal is to provide a new analysis tool for the DOB-controlled system based on nonlinear control theory, and its applications to a wide class of control systems.

Another remarkable issue of modern control system is to keep its state variables to remain inside a "safe region" during its operation. Such a feature is called "safety", which has received a lot of attentions in control and robotics society as a wide class of control systems are operating with a possibility of losing the safety (e.g., collision of two robots). We in CDSL@UoS are working on developing a control scheme that guarantees safety. A particular emphasis is placed on robustness of the safety notions against disturbance and model uncertainty, by which the control system may easily escape the pre-defined safe region.

As modern control systems are controlled by digital controllers and the control input is transmitted through a data network, the systems are sometimes interpreted as cyber-physical systems (CPS). When it comes to interpreting control systems in the CPS framework, a huge interest of research has been paid to the fact that the CPSs are exposed to threats of malicious agents or adversary outside the control loop, because of accessibility of data network. For those reasons, it is important to understand the vulnerability of a system in a control-theoretic point of view, and to exploit countermeasures to guarantee security of the system.

A recent approach of constructing a controller is to employ the redundant input/output data of the system instead of the system model identified in a traditional way, which is termed "data-driven control". Such a philosophy is sometimes beneficial when we have a huge amount of system data and the system order is too large. CDSL@UoS is focusing on revealing the underlying principle of this new technique and its applications.

Most of the legged robots encounter the balancing issue, which means that the robot should keep its postural stability while it completes task. Unfortunately, the balance could be easily lost when the robot is pushed by an unexpected force or it is walking/driving on an uneven terrain. CDSL@UoS is working for providing control-theoretic solutions to tackle this issue.

Sometimes it is not easy for single robot to complete a task in uncertain/complex environment. This is why the cooperative manipulation has received a huge amount of attentions in the literature. The problem turns out to be more challenging when the robots are "heterogeneous"; that is, each robot has different structure and capability, and the mobility and balance issue come into the picture, for which CDSL@UoS is trying to developing a yet another approach to cooperative manipulation.