The Automatic Control Laboratory of the Swiss Federal Institute of Technology (ETH), headed by Prof.~Manfred Morari consists of about 35 staff members. Its mission is the development of control theory, its application in various industries and to biomedical systems, and the development of the appropriate software tools.

About half the lab is dedicated to developing, extending and applying the latest theoretical methods of control. The main interests include modelling and identification, analysis, and control of hybrid systems; model predictive control; and computational issues associated with control. The other half of the lab is focused on applications, primarily in the area of biomedical engineering. The objective of the rehabilitation engineering group is to develop walking and grasping neuro-prostheses for subjects with stroke and spinal cord injury. The projects of the anesthesia control group encompass everything from the identification of dynamic models describing the effects of anesthetic drugs on humans, all the way to controller design, pilot studies and clinical validation (almost 200 operations with anesthesia under closed loop control).

Some activities in the hybrid systems area are detailed in the following. The group has developed a framework for modeling and controlling discrete time linear hybrid systems described by interdependent physical laws, logic rules, and operating constraints, denoted as Mixed Logical Dynamical (MLD) systems. In this framework the time evolution of the hybrid system is described by a set of linear equalities and inequalities involving both continuous and integer variables. While some of the system theoretic properties of this formalism remain to be fully explored, the approach is supported by the fact that it has been able to address effectively a number of long-standing problems. First of all, it has proven capable of describing many real engineering problems. More importantly, it has led to numerical (mathematical programming) approaches for solving these problems efficiently. The modeling effort is aided by a new language (HYSDEL) to describe hybrid systems in a compact way. The HYSDEL compiler generates automatically MLD models for control, estimation, and verification.

The system theoretic and engineering problems include identification, feedback control, state estimation and fault detection, verification, reachability, controllability, and observability. In all cases the analysis or synthesis problem is transformed to a mixed integer linear or quadratic program for which commercial software is available. For example, this approach allows one to find for a linear hybrid system the piecewise affine feedback controller which asymptotically stabilizes the system and optimizes some user specified objective function subject to constraints. An analgesia/anesthesia controller designed with this methodology was applied on five patients in the operating room. Among the many other case studies, where this approach was tested are a gas supply system at Kawasaki Steel, a dry-clutch, an automatic gear shift schedule for a Renault automobile, a traction control system for Ford, and a power management system for ABB.