Jan A. van Alste

Modelling the optimal control of cyclical leg movements induced by functional electrical stimulation

An optimal control strategy for FES-induced cyclical leg movements in paraplegics is proposed. The control of the cyclical movement of a freely swinging leg is considered as an example. Quadriceps and the flexion withdrawal reflex are stimulated in order to generate a cyclical movement, of which the forward swing resembles the swing phase of gait. Optimal stimulation patterns are determined on the basis of an optimization criterion and a dynamic model of the system. The criterion is based on desired movement parameters and a minimal duration of the stimulation bursts. The movement parameters should ensure the generation of the desired cyclical movement: a desired hip angle range, sufficient foot clearance during the forward swing and knee extension at the beginning of the backward swing. Minimal duration of the stimulation bursts is assumed to yield minimal fatigue. A dynamic model, describing the dynamics of the neural system, the muscles and the leg, was constructed and its parameters identified on the basis of preliminary experiments and literature. Optimal timing of the quadriceps and flexion reflex stimulation bursts was determined by means of computer simulation. These simulations predicted that the flexion reflex should be stimulated in a short burst approximately 150 ms before the start of the forward swing. The quadriceps should be stimulated approximately starting 200 ms before the end of the forward swing in order to ensure knee extension at the beginning of the backward swing. The duration of one cycle of the movement was between 1300 and 1500 ms in these simulations. These results predict that the movement specified by the functional objectives can be realised using only two channels of stimulation. On the basis of the optimal timing, an adaptive control strategy can be designed, which varies the stimulation burst width when muscles fatigue.

Dynamic pumping performance of the Hemopump - a small intraventricular blood pump

We studied the pumping characteristics of the Hemopump®, a commercially availabe miniature intraventricular blood pump for temporary support of failing hearts. The Hemopump® is an axial flow pump of which the characteristics can be described by turbomachine theory. Experiments with water and a mock circulation verified that the pumping characteristics of the Hemopump®, in terms of both pressure head and flow as a function of rotational speed, very well can be described by a first order differential equation. The influence of blood with its non-Newtonian character is being investigated

Dynamic pumping characteristics of the Hemopump

While pumping blood with the Hemopump® in sheep, the ability of predicting the instantaneous pump flow from the pressure difference over the pump system and pump parameters was investigated. For rotational speed n between 300 and 475 revolutions per second (rps), maximum pump flow QO(n) at zero pressure difference, internal pump resistance R(n), and inertia parameter Lc were found to be suitable parameters for Hemopump® characterization. The instantaneous pump flow could be estimated with an accuracy of approximately 1.0 [ml/s]. The values of the pump source parameters (± sd) were: (the figures in parentheses represent earlier reported values found while pumping water) Lc was a constant of 21.4 ± 6.4 [Pa·s2/ml] (in water: 10.8). QO(n) is linearly related to rotational speed n according to: QO(n) = Qo(ncen) + CQ(n - ncen), with QO(ncen) = 49.4 ± 4.5 [ml/s] (in water: 60.3), CQ = 142 ± 22.4 [10−3 ml] (in water: 146), and ncen = 387.5 [rps]. R(n) is linearly related to rotational speed n according to: R(n) = R(ncen) + CR(n - ncen), with R(ncen) = 556 ± 124 [Pa·s/ml] (in water: 502) and CR = 1.47 ± 0.83 [Pa·s2/ml] (in water: 1.67).

Finite State model in Control of Paraplegic Gait

Abstract A finite-state model of paraplegic gait may serve as the basis for the design of an hierarchical control system for locomotion assisted by Functional Electrical Stimulation (FES). The control system detects different states on-line during gait based on a model. In certain states, stimulation is applied. Experimental results of two state detection strategies are presented. The preliminary results include various sensory feedback signals, such as hip angle, footswitch and crutch force signals. Video recordings provided off-line comparison. State detection on the basis of footswitch and gonio data was problematic, probably due to time varying weight distribution in combination with a switch forcelevel and problems with robust attachment of the footswitches. Gonio together with crutch force data appeared to be more reliable for state detection. The latter sensor set was succesfully incorporated in an FES system for the finite-state control of paraplegic gait.