Testing and Evaluating the SKDS System

111

only visible by its corresponding visual results, or lack thereof. This leads directly to the next problem: feedback presentation. It is very problematic when the only feedback being generated is visual. While performing a gesture, the user often does not get any accompanying information aside from the perceptual haptic afferentation about the success or failure of their actions, and it is not possible to assist the user in correcting a gesture midway through. Hence, the user does not know if the right gesture was completed or where the gesture started and ended in a proper manner.

The use of tangential forces (real-time kinesthetic feedback) applied to a stylus tip allows us to support or even to augment perceptual haptic afferentation on the rigid surface of interactive displays, which solves the two most problematic issues in gestural interaction. Therefore, in our future work we plan to develop the SKDS prototype further, incorporating gestural interaction in a haptic space, actively guiding users in performing contextual gestures and assisting them, through force feedback, to complete or learn new input / interaction behavior.

6.5 T

E

SKDS S

To evaluate the SKDS system we employed a two prone approach. We first tested the various actuation components that would be ideal for creating directional forces on the Microsoft surface display. This was done by installing each component into the SKDS system and measuring its performance in an unloaded environment. Using this technique we monitored the speed, displacement, power efficiency and signal distortion of each component in the system. We then conducted a short user study to evaluate each actuator in a real-time loaded environment where users were asked to interact with the system to complete a basic drawing task.

Actuator Evaluation

The aim of evaluating the actuators was to understand and catalog their efficiency in generating enough directional forces in the SKDS setup, to observe the necessary stick-slip motion. As all the components being evaluated were not designed for this purpose it was imperative that the evaluation be done within the SKDS setup and the technical specification listed by the vendors only be used as a reference not as comparative measure. Secondly, it was also understood that as the components were not designed or optimized for the particular use case, their construction or methodology of generating vibrotactile feedback, may hinder their efficiency to generate stick-slip motion. Using this premise we recorded and evaluated various actuators performances and published our finding in IEEE and ACM conferences and journals (Farooq at al., 2016e; Farooq et al., 2017).

……………

112

Usability Testing

Using the technical setup which performed the best in the actuator testing scenario, we conducted various user studies to gauge the advantages of incorporating kinesthetic feedback through stick-slip motion. Recruiting university students (ages 29-34) to perform a set of drawing tasks on the SKDS with and without kinesthetic real-time feedback we evaluated the possible advantages of our feedback technique and the efficiency of the system. The participants were divided into 2 groups. Both groups were asked to draw four shapes (a rectangle, a square, a circle and an equilateral triangle) 5 times each, with haptic (kinesthetic) feedback and 5 times in the absence haptic feedback, using the SKDS. Using this methodology we recorded user performance and variations in user behavior along with their preferences and published these finding in IEEE and ACM conferences and journals (Farooq at al., 2016c; Farooq et al., 2017).

A part from the above mentioned research, a follow up study conducted in Aug 2017 (currently under review at SENSORS Journal), illustrates that although the current implementation of the SKDS system is sensible and greatly improves usability of any touchscreen / stylus based system, further enhancement could drastically improve usability and ease of use for a wide range of user activities. The yet unpublished study in question, invited 8 university students (ages 29-34) to perform simple drawing tasks of drawing four shapes (a rectangle, a square, a circle and an equilateral triangle) five times, on the SKDS system.

Each participant was asked to conduct this task with both their hands (dominant and non-dominant hand) in the absence and presence of haptic (kinesthetic) feedback. A calibrated algorithm approximated a reference point for each task and compared the results to the participants’ original task performance (dominant hand with no haptic feedback). The algorithm considered angle of vertices, slope of vertices and the orientation of the shape (with reference to the x-axis), to approximate congruency between each shape being drawn. Using this method a percentage deviation was calculated, with reference to the baseline (Fig.

51a & b). Using this method we evaluated performance differences between the actuators as well as the difference between each hand, limiting the effect of natural handwriting variations (Srihari, et al., 2001 and Williams et al., 2006) the participants may have, while conducting the drawing tasks.

……………

113 (a) (b)

Figure 51. Error percentage variations for each task across the use of all three actuators for dominant hand (a) and non-dominant hand (b).

To evaluate how much it affected each task (drawing the 4 different shapes), we compared kinesthetic feedback (solenoid based SKDS) with no haptic feedback, while the participants completed the task using the non-dominant hand (Fig. 52). Looking at the error percentages it is obvious that the SKDS improved task performance considerably.

Figure 52. Error percentage variations for non-dominant hand with and without haptic feedback.

Overall the error deviation of the non-dominant hand with haptic feedback remained just below 6% for all participants in all drawing tasks.

However, the same error deviation jumped to a maximum of 22%, with a minimum of 5% without haptic feedback. This shows that real-time kinesthetic feedback can greatly improve performance in a challenging task, such as drawing accurately with the non-dominant hand. However, as mentioned earlier, improvement in the design and actuation technology of the SKDS can increase the effectiveness of the system and may even restrict the user (through force feedback), in making these error in the first place. That is the true goal of the SKDS system; to function as an active I/O channel, guiding the user seamlessly through a given task. To achieve this we are continuing to develop our methodology and the design of the

……………

114

SKDS system as well as the application scenarios¹³ (see Ice Hockey Application Demo) in which the system can be utilized.

13 UIST2016 Ice Hockey Application Demo: https://youtu.be/-hXT96BgbZs

……………

115

7 A Short Summary of the