Interpretation of TactileGlove Data in Applied Use Cases
An App Note on Getting the Best from TactileGlove
Executive Summary
This white paper explains how to interpret PPS TactileGlove sensor data in simple gripping and lifting applications detailing: expectations, performance, mitigating error sources, and interpreting the correct force/pressure readings.
TactileGlove is a cutting-edge ergonomic testing tool that provides accurate and repeatable pressure measurements of the hand when interacting with the environment. Its high sample rate, repeatability, and sleek fit make it an indispensable tool for quantifying comfort and fit, as well as assessing effort and safety margins in tool and process development.
Quantification of tactile interactions is becoming the new standard for ergonomic testing, allowing designers to transition from subjective qualitative feelings into robust quantified sensations that allow for informed design decisions and optimized ergonomic products with fewer design iterations.
Quantified tactile data, being relatively new, can be difficult to interpret, and so this paper describes definitively how to reliably measure and interpret hand pressure data in a simple use case of lifting a dumbbell. This explores expectations, interpretations, and best practices to achieve robust and reliable ergonomic measurements time after time.
This paper adds confidence in measured tactile data using TactileGlove, enabling reliable measurements in customer applications and subsequent robust and informed design decisions that will shape ergonomic design now and in the future.
What is TactileGlove, and why is it used?
The PPS TactileGloves are tactile pressure sensor gloves with embedded sensing elements throughout the palm and fingers, allowing for natural and accurate testing of pressures applied to and exerted by the hand. Nearly any hand operation can be measured accurately through the 65 individual highly sensitive sensing elements within each glove. TactileGloves are wireless and battery powered, allowing for unencumbered movement and a natural use flow. This is the only glove sensor on the market that both allows for full range of motion and creates an organic testing environment. High-resolution pressure sensing and mapping enables you to research hand movements when performing various tasks, and analyze comfort, and ergonomics through visual feedback of hand interactions as shown.
TactileGlove is commonly used in the design and development of tools/equipment for long term use, requiring maximum comfort for reducing operator stress and fatigue. It is similarly used in research to quantify tactile sensations, as well as learning ‘skilled hand’ techniques in medicine, sports, and the arts.
Any hand based application or design task where operator sensation subjectivity is an issue, or where quantified data helps to monitor performance or compliance, TactileGlove is available to help designers and users understand tactile interactions. TactileGlove moves designers from an abstract qualitative world, to a definitive quantitative world where quantification leads to quality outcomes…
Theory of Operation
TactileGlove, like most other PPS products, consist of flexible electrode parallel plate capacitors where the output capacitance change is related to the applied pressure. The TactileGlove calibration turns the measured capacitance into the applied pressure. This can then be integrated over the electrode area in order to estimate applied force, if desired.
Nominally a pressure transducer, a scalar pressure distribution on the sensor electrode causes a deformation of the elastic dielectric proportional to the average pressure on the sensor element. The integral of the pressure field over a surface area yields the total force on that area of the sensor, as shown in the image. If this is done on the entire sensor then this yields the total force, a directional vector quantity oriented normal to the surface. If done over a smaller area, this yields multiple point loads centred on the area of interest. In typical cases of a flat electrode, these can be summed to give the total force on the sensor, however if the surface is curved then this must be done as a vector addition in order to result in the correct force.
TactileGlove converts capacitance into average pressure on the sensor element, and assumes this acts uniformly across the element area when performing integration. Similarly, when estimating total force on the glove, it assumes all elements are collinear. When looking at the TactileGlove, there are 2 areas of interest as shown:
A contact area where there is insensitive sensor material between the hand and tactile object
A smaller sensitive area that TactileGlove can read to determine pressure and loads. It is this area that is used for converting pressure to load.
Only pressure applied to, or loads transferred through, the active sensitive area will be reported by TactileGlove. This may not be the same as the contact area with the object of interest. The implications of this in interpreting tactile results are discussed in this white paper.
Calibration and Verification
Calibration Method
TactileGlove is robustly calibrated in a test chamber using an air bladder allowing for excellent linearity and repeatability over their rated ranges. The gloves are calibrated flat against a rigid plate which allows the bladder to compress the elements uniformly during calibration. Pressure is then increased in steps and piecewise functions provide a detailed calibrated response. This is done under controlled temperature and humidity conditions, with a minimally stretching bladder allowing for confidence and repeatability in sensor/chamber performance between calibrations, resulting in excellent linearity error of <2% with repeatability within 3% of the sensor rating, and a SNR of at least 500:1.
Trusting the Calibration
The calibration is expected to be valid for the life of the TactileGlove system, and can be sanity checked using a simple weighing scale, or load cell if you have one. Best done with a matching pad to ensure load is transferred through a particular element of interest, this method can provide confidence in the calibration when done carefully. For the case of loading a single element cluster onto a load cell/scale with a matching pad, the agreement between the glove and the reference is within the specified accuracy. This method is not suitable for checking the glove as a whole, as detailed in the following error sources section. When checking any individual element, it is important to tare (Zero) the glove in the test position (unloaded), meaning the hand is bent into the use position, allowing for residuals to be tared properly.
If the sensor is undamaged, but still not reporting expected values when doing a simple test like this, it may be explainable through one of the following 3 main error sources for TactileGlove as detailed in the following section, or require application specific interpretation as detailed in the applied example.
General Error Sources and Mitigation
Using the wrong size glove
Using the wrong size TactileGlove can cause excessive wear on the system, by allowing the glove to move (too large) or stretching (too small). Gloves that are too small are hard to get on and off, increasing the risk of damage. Gloves that are too large do not feel normal, and so may not give a natural tactile experience. On top of this, use of an improperly sized TactileGlove can have major impacts on the output accuracy. The image shows how this can happen. Say for some size (M), the pressure sensitive elements are located at optimal points on the hand for that size (the center of distal phalanx in the example). Other sized gloves used on this hand would put their respective sensor elements in suboptimal positions. A glove that is too small is the worst case, where the sensor would move towards the distal interphalangeal joint (in this case) which can cause unwanted flexure of the sensor and report inaccurate values. A glove too large will move the sensor elements towards the end of the finger, reducing the contact area between the finger and the elements, thus reporting inaccurate values. This is true for all elements, not just that illustrated. This is mitigated by a properly sized glove, using the sizing chart.
Temperature/Environmental Effects
TactileGlove calibration is robust to a wide range of environmental factors. However, interactions outside of the normal body/room temperature range such as: holding a hot cup of tea, or working outside in winter; will reduce the accuracy of the measurements by the temperature coefficient. This can be mitigated by using the device in the recommended operating range, or using suitable environmental control or insulation between the glove and the object of interest, also by algorithmic mitigation in post processing, however this can be cumbersome, so insulation is the preferred method when operating in extreme temperature ranges (0-50’C).
Improper Tare Position
Taring (zeroing out) the TactileGlove is important for controlling creep and residuals. Residuals can be created by the flexion of other sensor elements when loading, or moving into position to load the sensors.
Taring the sensors when in a representative position can reduce error significantly. The chart shows the effect of this in a cylinder grip example without the influence of external load.
Contact Mechanics
Somewhat related to the glove sizing is making sure the sensor elements are engaged with the object of interest, and that the finger/hand is engaged with the sensor elements. When the elements are fully engaged with a tactile feature of at least the size of the element, then the glove reports well, but below this accuracy derates non-linearly with the feature size (contact area).
Consider a simple case of a single element being loaded by a simple geometric shape as shown. Each of these cases will have a different contact area with the element and will respond differently under load, via deformation of the sensor electrodes. At low load, the elements are not significantly impacted, and so all cases behave normally, but at high loads the electrodes deform and this derates the performance.
This can be caused by tactile features in the application of interest, or in the case of shape ‘C’ can be caused by the finger (or other part of the hand) through improper sizing of the glove as the contact area between the hand and the element is altered. This can also be caused by bunching of the glove material, although this is rare.
This can be mitigated by the use of load spreading materials on the glove, or between the glove and the object, ensuring the glove material is flat over the elements, or by applying a post-processing correction based on empirical measurements of derating vs feature size. This correction will be device dependent, as the relation between one hand, a particular size of glove, and the tactile object is use case specific.
The above example deals with making sure the applied load is spread across the sensing element, the following situation deals with making sure the applied load goes through the element at all. In some cases force is applied to an object where there is no tactile element (called a force short), and so the total load on the object will not be reported correctly by the TactileGlove, although the glove will correctly report/ measure the force/pressure at the element locations.
This is not something that necessarily needs ‘mitigating’ as such, as the gloves are correctly measuring the forces at that point of the hand, which is not always the same as the total force applied to the object. However, if exact values of total force are required, it will be necessary to reduce load shorts using spacer pads on the tactile elements (so only the tactile elements are in contact with the object), or by integrating the measured element pressures over the remaining hand contact area (determined empirically) in post processing. This matter is subtle, yet important, and will be dealt with more thoroughly in the following worked application example.
Applied Application Interpretation
Worked Application: Lifting a Dumbbell
A good TactileGlove application example for analysis, related to tool design, sports equipment design, and sports science, but also generally applicable, is measurement of lifting a dumbbell.
With regards to the measurement error sources listed above, this is an application where: Loads are well within the TactileGlove rating, it uses smooth geometry (at the element size scale), it has multiple use conditions that highlight some potential traps, and is simple to follow and visualize the forces involved allowing conclusions to be drawn.
Considering the image below, a 4.5kg plastic/concrete dumbbell is lifted and held without movement in 3 different cases, which are annotated with observations:
Weight resting on the fingers only
Finger elements are well engaged with the object, and so total load from TactileGlove does yield the object mass correctly!
It can be observed that most load is taken by the index and middle fingers, but considerable load is on the ring finger too.
Hand wrapped gently around (some palm area)
The finger elements are well engaged, but the object has rolled back onto the palm, touching an area without tactile elements, resulting in total object mass reported being smaller than the true mass. Loads on the fingers/elements are still correct, and are showing a load redistribution, with most load now on the index finger, with the rest transferred to the palm. The reduced load on the ring finger suggests this is more comfortable (it certainly was!)
Hand firmly gripping the dumbbell
Here the elements are well engaged, with some areas without tactile elements, but now the load reported does not correspond with the true object mass due to the grip force.
All elements show increased load, but it is well distributed and so still comfortable. In this case it is element pressures that indicate comfort and fit, rather than total load. Object mass can be backed out of this measurement, but it requires a free body diagram (FBD)
Interpreting Lifting Forces
In condition 1, where the load is transferred directly through the elements, without load shorts, the interpretation is simple as the loads sum to the correct value allowing determination of both the finger pressures and the object mass.
In condition 2, where the load is transferred through the elements and some insensitive areas (with load shorts) interpretation is more complex.
So in this case, the finger loads (and palm too for that matter) are reporting correctly as they were tared and verified before hand, and so conclusions about localised pressures (or localised forces) can be safely drawn. In this case we have a lowering of finger loads as some is transferred to the palm, and also a redistribution of the finger loads showing that the mass has shifted and is being held differently, relating it to comfort (from the redistribution I can correlate my sensation of this being better to a quantified load and finger distribution).
As the lateral forces in this situation are negligible (there is minimal squeezing), and the reference load of 4.5kg is known, the load difference can be used to provide a crude approximation of the insensitive area (or total contact area) which can help to return the object mass if desired in later directly comparable experiments.
Interpreting Grasping Forces
Condition 3 is the most complex example, but will be representative of many customer applications, with both a static load (the object mass) and grip/grasp forces.
To start off, if looking at individual element loads/pressures, these will be reporting correctly and can be easily compared to other situations and interpreted. In this case there is a general increase in load across all digits, but without significant redistribution suggesting that the object is held similarly to condition 2, but with more grip strength. It also shows the inclusion of the thumb now as the thumb wraps around the top of the bar. This is useful for setting a baseline of comfort for this particular dumbbell, for comfort comparisons to others, or for visualizing changes in grip distribution over time or between people. This is the type of analysis that TactileGlove excels in, allowing for quantification of sensations and measurement of subtle changes over time or between experiments.
With additional context for this application, beyond what TactileGlove can currently report, specifically the positions of the tactile elements on the object of interest (the dumbbell), the total object mass and directional loads can be determined from the subsequent FBD. TactileGlove is fundamentally a pressure measurement tool, and gets forces by integrating pressure over the element area, and so does not handle the vector nature of forces well without the customer/user context of the forces. As stated, with this context, highly accurate measurements of both localised forces/pressures (important for comfort and fit) and total vectorised forces (important for measuring effort, work, and unknown object masses) can be made. In most applications however, it will be the simply interpretable localised loads/pressures that are important.
Best Practices
1. Ensure glove is correctly sized
As shown, the use of a properly sized glove is important to maintaining accuracy in the system, as well as improving the lifespan. It can be tempting to use whatever glove is available for a variety of users, however best results will be achieved with properly sized gloves for each hand size/type, where issues relating to unusual loading are minimized.
2. Ensure Glove is seated correctly
Along with the sizing, ensuring the elements are correctly seated in their respective locations is important both for accuracy, and to provide representative outputs. For example if your application calls for loads at the side of the fingers (where elements are typically on the inside of the finger), the elements can be rotated around slightly or shifted minutely up or down so they are engaged with your application.
3. Prepare for application environmental conditions
If the application involves water or solvents, then appropriate outer gloves should be used over TactileGlove. This may restrict the maximum single test time, to reduce the sweat build up in the TactileGlove. Similarly, if being used in temperatures outside of normal laboratory conditions, or in applications normally requiring gloves, appropriately sized insulation/gloves should be used over TactileGlove (sized to fit over TactileGlove relatively loosely).
4. Taring in right position
For general applications,TactileGlove should be tared (zeroed) in a neutral hand position as shown. However, if detailed measurements of a particular action are being performed, best accuracy can be achieved if TactileGlove is tared with the hand in a similar but unloaded position to the intended application. If measuring grasping forces on a cylinder, the TactileGlove should be tared with the fingers curled at approximately the cylinder size, prior to measurement.
5. Load spreading
TactileGlove performs best with smooth geometries (in the scale of the sensor elements). When dealing with tight features smaller than the element size, unless empirical corrections are available, a semi-rigid load spreading material will result in best load accuracy. The stiffness/thickness of the material depends on the load and feature size. At low loads 2mm of soft rubber will be sufficient, but at 100% of the rating with a ‘sharp’ feature, this will need to be increased or made stiffer (leather/plastic/wood).
6. Determine what is important in results interpretation (total applied load, or local load)
Determine what it is that you are trying to measure, as it will help with understanding results and preparing a representative experiment, it will also set realistic expectations on what the TactileGlove will report. If you require total applied load accurately, then considerations for load spreading and element separation must be made. If you are looking at comfort and fit, localised loads/pressures will be best and so it may be that the total load values from the glove are not what is important, focusing on loads at individual points instead
Useful Reading and Support
Additional white papers for calibration/verification and application support can be found here:
WHITE PAPER - Quantifying Hand Ergonomics With The TactileGlove - Published 2021
WHITE PAPER - Calibrating and Verifying Tactile Pressure Sensors - Published 2018
Details on TactileGlove specifications and intended use applications can be found here:
Force Measuring Sensor Glove | Grip & Hand Mapping
Pressure Mapping & Pressure Mapping Technology
Real examples of TactileGlove being used in research and development can be found here:
RESEARCH ARTICLE - Investigating Gripping Force During Lifting Tasks Using a Pressure Sensing Glove System - 2月 2023 - Science Direct Applied Ergonomics - Download
RESEARCH ARTICLE - Identification of Adaptive Driving Style Preference through Implicit Inputs in SAE L2 Vehicles - 11月 2022 - Honda Research Institute USA, Inc - Download
RESEARCH ARTICLE (in Japanese) -Measurement and Assessment of Touch Skills during Dementia Care Movements Using Tactile Gloves - 2020 - Kyushu University
To find out more about out TactileGlove or get a quote, please contact PPS Sales representatives.
