Version 2 2020-02-03, 16:28Version 2 2020-02-03, 16:28
Version 1 2011-03-04, 09:11Version 1 2011-03-04, 09:11
thesis
posted on 2020-02-03, 16:28authored byMichael D. O'Toole
The treatment response for traumatic head injured patients can be improved by
using an autonomous robotic system to perform basic, time-critical emergency neurosurgery,
reducing costs and saving lives. In this thesis, a concept for a neurosurgical robotic system is proposed to perform three specific emergency neurosurgical procedures; they are the placement of an intracranial pressure monitor, external
ventricular drainage, and the evacuation of chronic subdural haematoma. The control
methods for this system are investigated following a curiosity led approach. Individual problems are interpreted in the widest sense and solutions posed that are general in nature. Three main contributions result from this approach: 1)
a clinical evidence based review of surgical robotics and a methodology to assist in their evaluation, 2) a new controller for soft-grasping of objects, and 3) new propositions and theorems for chatter suppression sliding mode controllers. These contributions directly assist in the design of the control system of the neurosurgical robot and, more broadly, impact other areas outside the narrow con nes of the target application. A methodology for applied research in surgical robotics is proposed. The methodology sets out a hierarchy of criteria consisting of three tiers, with the most important being the bottom tier and the least being the top tier. It is argued that
a robotic system must adhere to these criteria in order to achieve acceptability. Recent commercial systems are reviewed against these criteria, and are found to conform up to at least the bottom and intermediate tiers. However, the lack of
conformity to the criteria in the top tier, combined with the inability to conclusively
prove increased clinical benefit, particularly symptomatic benefit, is shown to be hampering the potential of surgical robotics in gaining wide establishment. A control scheme for soft-grasping objects is presented. Grasping a soft or fragile object requires the use of minimum contact force to prevent damage or deformation. Without precise knowledge of object parameters, real-time feedback
control must be used to regulate the contact force and prevent slip. Moreover, the controller must be designed to have good performance characteristics to rapidly modulate the fingertip contact force in response to a slip event. A fuzzy sliding mode controller combined with a disturbance observer is proposed for contact force control and slip prevention. The robustness of the controller is evaluated through
both simulation and experiment. The control scheme was found to be effective and robust to parameter uncertainty. When tested on a real system, however, chattering phenomena, well known to sliding mode research, was induced by the
unmodelled suboptimal components of the system (filtering, backlash, and time delays). This reduced the controller performance. The problem of chattering and potential solutions are explored. Real systems using sliding mode controllers, such as the control scheme for soft-grasping, have a tendency to chatter at high frequencies. This is caused by the sliding mode
controller interacting with un-modelled parasitic dynamics at the actuator-input
and sensor-output of the plant. As a result, new chatter-suppression sliding mode controllers have been developed, which introduce new parameters into the system. However, the effect any particular choice of parameters has on system performance
is unclear, and this can make tuning the parameters to meet a set of performance
criteria di cult. In this thesis, common chatter-suppression sliding mode control
strategies are surveyed and simple design and estimation methods are proposed.
The estimation methods predict convergence, chattering amplitude, settling time,
and maximum output bounds (overshoot) using harmonic linearizations and invariant
ellipsoid sets.
History
School
Mechanical, Electrical and Manufacturing Engineering