Mathematically rigorous techniques for motion-planning and control of robots with which direct control of every degree of freedom is impossible, e.g. a biped walker with point feet or a nonholonomic wheeled robot. The governing principle is that a system's natural dynamical behaviour should not be overcome, but used.
The central technique is to define a target motion, not as a function of time, but as a submanifold of the state space on which the natural dynamics admit (partial) closed-form solutions. Controllers are then defined to stabilize dynamics transverse to this manifold.
[More Info] External partners: Russ Tedrake, Andrey Savkin
Combining first-principles modelling and modern system identification techniques to analyze the dynamics of the human cerebrospinal fluid system. Development of information-optimal diagnostic procedures to distinguish healthy patients from those who would benefit from surgery, with particular attention to the diagnosis of hydrocephalus.
[More Info] External partners: Umeå Hydrocephalus Group, Likvor AB
Industry-linked project to develop advanced control and teleoperation systems for hydraulic forestry cranes, with a long-term vision towards completely autonomous operation.
External partners: Komatsu Forest AB (Valmet), Cranab AB, IFOR
Biography
Ian Manchester was born in 1979 in Sydney, Australia. His PhD thesis was in the area of vision-based guidance and robot navigation, and introduced the method of Circular Navigation Guidance. His advisor was Professor Andrey Savkin.
Since 2006 he has been with control systems group at Umeå University, Sweden. In that time, he has been a visiting researcher at the University of New South Wales (Australia), Sungkyunkwan University (Republic of Korea), and the Massachusetts Institute of Technology (USA).
As of September 2009 he will be at the Robot Locomotion Group, Massachusetts Institute of Technology.
