Technologies

The most commonly used technologies for sensing forces are strain gauges, piezoelectric sensors, capacitive-based sensors, and optical sensors. These technologies and other more novel methods are summarized in Table 2.3 and discussed below.

Strain gauges: The most common technology employed for force sensing uses thin metal foils applied to the surface of an instrument to measure the deformation caused by the applied force. These foils, when attached to a thin plastic backing material and oriented in a particular arrangement, are called strain gauges [64]. Strain gauges have been successfully used in instruments for medical applications [23, 36, 65, 66] (more details on these instruments are provided in Section 2.4.2). Figure 2.4 shows examples of strain gauges applied to MIST instruments.

To achieve multi-axis measurements using strain gauges, special structural elements are com- monly used on which strain gauges are placed at different locations to allow the different forces and moments to be measured. A review of force sensing structures found in the literature is presented in [67]. The Stewart Platform [68], the Maltese Cross [69], and Junyich’s configuration [70, 71] and their variations are the most commonly used structures. Other novel structures for multi-axis sensing are presented in [67, 72–74]. The selection of one of these structures for force sensing purposes depends on the desired balance between signal noise levels, measurement isotropy, signal coupling, the number of sensing elements and the size of the structure.

2.3 Force Sensing 16

Table 2.3: Force sensing technologies.

Technology Advantages Limitations

Strain gauges

Small size and can be sealed in a waterproof environment. Multi-axis measurement is easily achieved.

Sensitive to electromagnetic noise and temperature changes leading to drift and hysteresis. Tradeoff between the sensitivity of the measurement and the stiffness of the structure [4].

Optical sensors

Forces can be measured in as many as 6 DOFs [75]. They can be used inside Magnetic Resonance Imaging (MRI) scanners [76]. Also, they can detect changes with high sensitivity and reproducibility with no hysteresis [75].

Limitations include sensitivity to noise, and that optical fibres cannot typically achieve small bending radii [4].

Measurement of actuator input

The system is no longer limited by the sensor bandwidth (which can make a control or feedback system unstable), and it is not necessary to incur the cost of force sensors [77]. Does not rely on force sensors, which often do not operate properly when exposed to high temperatures and humidity [78].

Very sensitive to uncertainties [78]; if the system cannot be properly modelled (due to high joint friction, for example), the estimation error can be significant.

Capacitive- based sensing

Limited hysteresis, better stability and increased sensitivity compared to strain gauges [4, 79].

Require more complex signal processing and are more expensive than other methods [4, 79]. Resonance-

based sensing

High signal to noise ratio and digital processing is

possible. Affected by nonlinearities and hysteresis [79].

Piezoelectric sensing

Since these materials generate their own voltage, no additional power supply is needed [4]. They have high bandwidth, high output force, compact size and high power density [79].

Very temperature dependent and subject to charge leakages [4]). This results in a drifting signal when static forces are applied, thus making them suitable for the measurement of dynamic loads only.

(a) (b)

Figure 2.4: Examples of strain gauges used in instruments to measure forces during natural orifice procedures [37] (left) and during laparoscopic procedures [80](right).

2.3 Force Sensing 17

Optical sensors: Another technology often used for sensing force is based on measuring the change in intensity or phase of a light signal as it passes through a flexible structure. When forces are applied to the structure, the way in which the structure flexes creates a change in the intensity of the light or causes the phase of the light signal to vary proportionally, making it possible to estimate the amount of force or pressure being applied. The design of 6-DOF optical force sensors is presented in [75,76], while the development of optical force sensors for MIST is discussed in [81,82]. Measurement of actuator input: When actuators are used to drive the joints of a manipu- lator, the input to the motor (for example, the current drawn by electrical motors or the variation in the pneumatic pressure in pneumatic actuators), can be directly related to the amount of force or torque generated. Since these signals are affected by internal joint friction and actuator nonlin- earities [4], knowledge of the mechanism kinematics and friction dynamics is critical for force and torque measurements to be accurate. Measurement of actuator input is mostly used in master–slave (teleoperation) systems for force or torque control or for haptic feedback.

In order to use the input to the actuators as a means of estimating applied forces, an observer- based control system is commonly used. For this purpose, a model of the system needs to be developed and the uncertainties of the model compensated for or measured. Observer-based sensing compares the difference between the output of the nominal model and the actual system output. If the uncertainties are known, the disturbance observer can estimate the amount of force generated at the output based on the actuator input [83]. To measure uncertainties, [84] uses neural networks to estimate friction, inertia, and gravity, while [78] uses a modified extended Kalman filter to compensate for the modelling error, sensing bias and measurement noise. A Nicosia state observer is utilized in [85] together with a general bilateral control law that ensures matching of the forces at the master and the slave. Examples of these types of controllers are presented in [77, 78, 83, 84], while only the latter has actually been used in MIST.

Capacitive-based sensing: This type of sensing is commonly used to measure tactile infor- mation. It depends on the use of a membrane, which when deflected, causes the distance between two electrodes separated by a dielectric material to change. Examples of instruments based on this concept include [86, 87].

Resonance-based sensing: This type of sensing is also membrane based. A change of force and pressure can be detected by measuring the change in the resonant frequency of the membrane. Piezoelectric sensing: In piezoelectric materials, a change in mechanical stress results in

2.3 Force Sensing 18

a voltage change across the material [79]. The most commonly used piezomaterials for sens- ing and actuation purposes include a piezoceramic called Lead Zirconate Titanate (PZT) and a piezopolymer called Polyvinylidene Fluoride (PVDF) (see [88] for more details on these and other piezomaterials).

Other technologies: It is possible to find in the literature other technologies for measuring forces that still have not been used in MIST applications. Deflection sensors are one of these technologies, which are based on the ability to measure the deflection of a component with known material properties to estimate the amount of force being applied on it. The key for these sensors to be effective is to be able to properly measure displacement. Displacement sensors include potentiometers, Linear Variable Differential Transformers (LVDTs), and encoders [89]. A 3-DOF force sensor based on measuring beam deflection using LVDTs is presented in [90].

Some force sensors utilize piezomagneticmaterials, in which a change in the stress applied can be detected as a change of permeability in the presence of a magnetic field [79]. In [91], a 6-DOF force sensor has been designed usingultrasound transducers. The sensor measures the amount of time it takes for an ultrasound pulse to travel from the emitter to the sensor. As forces are applied to the object, an elastomer layer deforms, changing the distance that the ultrasound pulse needs to travel. A series of transducers are placed in a particular pattern to measure multi-axial forces and torques. The authors claim that these sensors are highly accurate, robust and inexpensive compared to other modalities.