Monitoring Skin Temperature during Hand-Object Interactions

Hsin-Ni Ho^1,2 and Lynette A. Jones^{1 1}Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA ²NTT Communication Science Laboratories, NTT Corporation, Atsugi, Kanagawa, Japan

ABSTRACT

An IR thermal measurement system was created to monitor skin temperature during hand-object interactions. The layout and optical arrangement of the system allowed for the measurement of skin temperature across the fingerpad during contact, together with contact force and contact area. An IR model was developed to compensate for the effects of the optical components that were placed between the fingerpad and IR camera. Based on this model, the measured thermal energy could be converted into a temperature distribution on the fingerpad during contact. The performance of this system was evaluated with calibration tests and the results indicated that this system was capable of providing accurate temperature measurements without the limitations imposed by conventional thermal sensors. Measurements obtained with this system can be used to evaluate the validity of models used to predict the changes in skin temperature during hand-object interactions.

INTRODUCTION

Modeling the thermal responses that occur when the hand makes contact with an object is essential for developing thermal feedback systems for haptic displays. When evaluating a thermal model from a physiological perspective, the temperature responses of the skin during contact are typically measured with small thermal sensors, such as thermistors or thermocouples affixed to the skin. However, it is difficult to obtain consistent temperature measurements with sensors. When the sensor is placed directly on the contact area [1, 2], it molds to the fingerpad and affects the surface area between the fingerpad and material. When it is attached to the perimeter of the contact area [3, 4], the localized nature of the change in skin temperature means that the sensor is not able to detect the full extent of the temperature change during contact. As a result, there are considerable differences in the magnitude of the changes in skin temperature recorded as a function of the location of the thermal sensor, with larger decreases reported when the sensor is directly on the contact area as shown in Figure 1. In addition, the measured time courses and amplitude of the skin temperature changes during contact are significantly different from theoretical predictions [5].

SS

40

35

30

25

20

Time (s)

Figure 1. Changes in skin temperature during contact with stainless steel with the sensor attached to the center (dashed line) or perimeter (dotted line) of the contact area. Model predictions from Ho and Jones [5] are also shown (solid line).

Skin temperature (^oC)

It is unclear whether the discrepancies between the predicted and measured changes in skin temperature reflect the limitations imposed by thermal sensors or inadequacies in the thermal model. A clearer understanding of the thermal response of the skin to contact would be possible if a non-contact thermal measurement system was used. As a result, an infrared thermal measurement system has been developed to provide an image of the temperature distribution across the fingerpad during contact. This system has been designed so that it can measure contact area, contact force and the change in skin temperature simultaneously during contact. These measurements permit the validation of thermal models, which can be incorporated into haptic displays to provide thermal feedback.

SYSTEM LAYOUT

When an infrared camera is used to measure skin temperature during contact, the optical components placed between the infrared camera and the fingerpad must be infrared transmissive to prevent the obstruction of the thermal radiation emitted from the fingerpad. The layout of the system designed to fulfill these requirements is shown in Figure 2. An infrared camera (A40M, FLIR Systems) was used to measure the thermal radiation emitted from the fingerpad. In order to measure the contact area and temperature distribution on the fingerpad simultaneously, a ZnS infrared window that can transmit wavelengths in both the visible and infrared spectrum was selected as the contact material. It was 43 mm in diameter and 10 mm thick. A beamsplitter, made of Ge with an anti-reflection coating, was used to separate the infrared radiation and visible light from the contact area. It measured 32 mm in diameter, was 3 mm thick and was held at a 45^o inclination. The contact area was captured by a digital camera (EOS D60, Canon). A 6-axis force transducer (Nano 43, ATI Industrial Automation) with a 20-mm diameter hole in the center was attached to the contact material and measured the contact force without obscuring the infrared and visible radiation from the fingerpad. During measurement, the distance between the fingerpad and the infrared camera was 300 mm, which was the minimum distance required for focusing. The overall setup of the infrared thermal measurement system is shown in Figure 2.

Figure 2. Layout of the infrared thermal measurement system.

PARAMETERS IN INFRARED MEASUREMENT

In this thermal measurement system, the radiation measured by the infrared camera not only originates from the fingerpad, but also from the surroundings and the optics placed in between the camera and finger. To measure the fingerpad temperature accurately, it was necessary to establish an IR model which compensated for the effects of the different radiation sources. In order to achieve this goal, the following optical parameters needed to be considered.

Target parameters The fingerpad was the target of this thermal measurement system. In this study, the fingerpad was a gray body with a constant emissivity of 0.95 [6] within the spectral range of the infrared camera, that is 7.5-13 µm. A calibration test was conducted with the system to validate this assumption which will be discussed later. Ambient parameters Together, the target distance, relative humidity, and ambient temperature determine the transmission of the air between the target and the infrared camera. The total transmissivity of the atmosphere over a distance is D =Tm x Ts, where Tm is the molecular absorption by constituent gases and Ts is scattering by particles in the atmosphere. In the present configuration, the target distance was 300 mm and therefore the transmissivity of the atmosphere equals 1 [7]. The ambient temperature determines the amount of thermal radiation from the surrounding sources.

External optics parameters The infrared camera selected for this system measures over the waveband of 7.5 to 13 µm. In order to characterize the optical properties of the external optics in this spectral range, that is the contact materials and beamsplitter, the spectral transmissivity and reflectivity of the optics were measured with a spectrometer (Nicolet Magna 860 Fourier Transform Infrared Spectrometer). The angles of incidence for the measurement were chosen to be the same as the system configuration, that is normal and 45^o for the contact material and beamsplitter, respectively.

The optics used in the system were real body radiators and exhibited a bandpass transmissivity that depended on the waveband response of the infrared camera, the spectral response of the optics, and the temperatures of the optics and the target. Based on the model proposed by Madding [8, 9] with the assumptions that the surfaces of the target and optics were diffuse and that the optics’ temperature equaled the ambient temperature, the bandpass transmissivity, τ , can be estimated from:

¹³ τ (λ) ⋅ R (λ) ⋅[E(λ,T ) − E(λ,T )]dλ

det tar amb

^∫7.5

τ=₁₃ (1)_∫ R_det(λ) ⋅[E(λ,T_tar ) − E(λ,T_amb)]dλ

7.5

where τ is transmissivity, Rdet is the detector response of the infrared camera, E is the spectral emissive power of a blackbody, λ is wavelength, and T is temperature in Kelvin. Subscripts tar and amb represent target and ambient, respectively.

The bandpass reflectivity ρ , is defined as the fraction of the spectral irradiation that is reflected by the surface [6]. With respect to the reflectivity of optics at the infrared camera side, the ambient radiation was assumed to be the only source of irradiation. With the assumption that the surface of the optics is diffuse, the bandpass reflectivity can be calculated from:

¹³ ρ(λ)E(λ,T_amb )dλ

∫7.5

ρ= ₁₃ (2)E(λ,T_amb )dλ

∫7.5

By working with the bandpass properties of the optics, the optics can be treated as gray bodies within the infrared camera’s spectral range. The bandpass emissivity, ε , can therefore be estimated from the relation

ε +τ +ρ = 1. For a typical hand-object interaction with the initial temperatures of the skin and object of 34 and 24 ^oC, respectively, the skin surface temperature after 10 seconds of contact should be around 27 ^oC based on the thermal model proposed by Ho and Jones [5]. The bandpass optical properties of this system were therefore evaluated based on the ambient and target temperature of 24 and 27 ^oC, respectively.

IR MODEL FOR TARGET TEMPERATURE DERIVATION

Figure 3. Schematic representation of the IR model.

In order to derive the temperature of the fingerpad during contact based on the thermal energy measured by the infrared camera, an IR model was developed to compensate for the effects of the contact material and beamsplitter placed between the fingerpad and IR camera. The schematic representation of the IR model is shown in Figure 3. This model takes into account various sources of thermal radiation that can be detected by the IR camera. They include the radiation emitted by the contact material, beamsplitter, and target, as well as the ambient radiation reflected by the contact material, beamsplitter and the target. Based on this model, the radiation emitted by the target can be derived as:

^Utot ^=εobj^τ1^τ 2^Uobj ^{+ (1−ε}obj ^)τ1^τ 2^Uamb ^+ε2^Uopt2 ^+ε1^τ2^Uopt1 ^+ρ1^τ2^Uamb ^+ρ2^Uamb

(3)

^Utot ^{− (1−ε}obj ^)τ1^τ2^Uamb ^−ε2^Uopt2 ^−ε1^τ 2^Uopt1 ^−ρ1^τ 2^Uamb ^−ρ2^Uamb

⇒ U =

obj