Abstract-This paper presents an analysis of the extended Kalman filter formulation of simultaneous localisation and mapping (EKF-SLAM). We show that the algorithm produces very optimistic estimates once the "true" uncertainty in vehicle heading exceeds a limit. This failure is subtle and cannot, in general, be detected without ground-truth, although a very inconsistent filter may exhibit observable symptoms, such as disproportionately large jumps in the vehicle pose update. Conventional solutions-adding stabilising noise, using an iterated EKF or unscented filter, etc-do not improve the situation. However, if "small" heading uncertainty is maintained, EKF-SLAM exhibits consistent behaviour over an extended time-period. Although the uncertainty estimate slowly becomes optimistic, inconsistency can be mitigated indefinitely by applying tactics such as batch updates or stabilising noise. The manageable degradation of small heading variance SLAM indicates the efficacy of submap methods for large-scale maps.

A clinically intuitive physiologic controller is desired to improve the interaction between implantable rotary blood pumps and the cardiovascular system. This controller should restore the Starling mechanism of the heart, thus preventing overpumping and underpumping scenarios plaguing their implementation. A linear Starling-like controller for pump flow which emulated the response of the natural left ventricle (LV) to changes in preload was then derived using pump flow pulsatility as the feedback variable. The controller could also adapt the control line gradient to accommodate longer-term changes in cardiovascular parameters, most importantly LV contractility which caused flow pulsatility to move outside predefined limits. To justify the choice of flow pulsatility, four different pulsatility measures (pump flow, speed, current, and pump head pressure) were investigated as possible surrogates for LV stroke work. Simulations using a validated numerical model were used to examine the relationships between LV stroke work and these measures. All were approximately linear (r(2) (mean ± SD) = 0.989 ± 0.013, n = 30) between the limits of ventricular suction and opening of the aortic valve. After aortic valve opening, the four measures differed greatly in sensitivity to further increases in LV stroke work. Pump flow pulsatility showed more correspondence with changes in LV stroke work before and after opening of the aortic valve and was least affected by changes in the LV and right ventricular (RV) contractility, blood volume, peripheral vascular resistance, and heart rate. The system (flow pulsatility) response to primary changes in pump flow was then demonstrated to be appropriate for stable control of the circulation. As medical practitioners have an instinctive understanding of the Starling curve, which is central to the synchronization of LV and RV outputs, the intuitiveness of the proposed Starling-like controller will promote acceptance and enable rational integration into patterns of hemodynamic management.

From the moment of creation to the moment of death, the heart works tirelessly to circulate blood, being a critical organ to sustain life. As a non-stopping pumping machine, it operates continuously to pump blood through our bodies to supply all cells with oxygen and necessary nutrients. When the heart fails, the supplement of blood to the body's organs to meet metabolic demands will deteriorate. The treatment of the participating causes is the ideal approach to treat heart failure (HF). As this often cannot be done effectively, the medical management of HF is a difficult challenge. Implantable rotary blood pumps (IRBPs) have the potential to become a viable long-term treatment option for bridging to heart transplantation or destination therapy. This increases the potential for the patients to leave the hospital and resume normal lives. Control of IRBPs is one of the most important design goals in providing long-term alternative treatment for HF patients. Over the years, many control algorithms including invasive and non-invasive techniques have been developed in the hope of physiologically and adaptively controlling left ventricular assist devices and thus avoiding such undesired pumping states as left ventricular collapse caused by suction. In this paper, we aim to provide a comprehensive review of the developments of control systems and techniques that have been applied to control IRBPs.

The application of rotary left ventricular (LV) assist devices (LVADs) is expanding from bridge to transplant, to destination and bridge to recovery therapy. Conventional constant speed LVAD controllers do not regulate flow according to preload, and can cause over/underpumping, leading to harmful ventricular suction or pulmonary edema, respectively. We implemented a novel adaptive controller which maintains a linear relationship between mean flow and flow pulsatility to imitate native Starling-like flow regulation which requires only the measurement of VAD flow. In vitro controller evaluation was conducted and the flow sensitivity was compared during simulations of postural change, pulmonary hypertension, and the transition from sleep to wake. The Starling-like controller's flow sensitivity to preload was measured as 0.39 L/min/mm Hg, 10 times greater than constant speed control (0.04 L/min/mm Hg). Constant speed control induced LV suction after sudden simulated pulmonary hypertension, whereas Starling-like control reduced mean flow from 4.14 to 3.58 L/min, maintaining safe support. From simulated sleep to wake, Starling-like control increased flow 2.93 to 4.11 L/min as a response to the increased residual LV pulsatility. The proposed controller has the potential to better match device outflow to patient demand in comparison with conventional constant speed control.