Modern life cannot function without automatic control systems, but they are often overlooked or badly designed. They range from the simple heating regulators in your home, through satellite, aircraft and underwater vehicle tracking controllers, and on to distributed control structures that run whole factories and power stations.
In ECS we have wide expertise that includes iterative learning control, repetitive control, behavioural theory for multidimensional systems, flow control, nonlinear control, adaptive control, robust stability theory, and data-driven control. Recent industrial collaborations have involved BAe Systems, Airbus, National Grid Transco, with further support from The Royal Society, the EU and EPSRC. We have a long history of control system design for rehabilitation engineering, industrial robotics, process control, underwater autonomous vehicles, wind turbines, power electronics and satellite control.
One of the most complex systems to model and control is the human body. In the BIO research group we have a strong track record in control systems design to assist human movement, specifically targeting conditions such as stroke, Multiple Sclerosis, Parkinson’s Disease and Spinal Cord Injury. Here we employ technologies such as conventional robotics, Functional Electrical Stimulation (which artificially activatesmuscles), and soft robotic structures to replace or augment users’ movement. We combine them with measurements provided by electromyography (EMG) to detect muscle activity, electroencephalogy (EEG) to detect brain activity, electrocardiography (ECG) to measure heart activity, non-contact depth sensors, and body-worn sensors to provide other information about the human body.
To design an accurate controller requires some form of mapping to be constructed between the sensor (input) data and the assistive (output) signals. An example of the blocks involved in an anthropomorphic system is shown below and includes principal inter-actions associated with assistive and rehabilitative technology.
Control system design must balance complexity, usability, set-up time, performance and the ability to change with the underlying human body. Our methodology therefore exploits structural knowledge, learning and adaption to developsophisticated automatic controllers.
One example of a control system we have developed is the next generation of wearable devices for people who have suffered a stroke, in the project SmartMOVE.
This helps stroke patients perform tasks such as eating, dressing and manipulating objects. Because it must work in homes with no engineer or clinician to set-up and tune it each day, it necessitates a brand new type of control structure that can learn from experience, adaptbetween multiple models of the user’s underlying dynamics, and have the ability to complement their remaining voluntary motion.