Performance and Comfort of Precise Distal Pointing Interaction in Intelligent Cockpits: The Role of Control Display Gain and Wrist Posture

Abstract

Using personal smart devices such as mobile phones to perform precise distal pointing in intelligent cockpits is a developing trend. The present study investigated the effects of different control display gains (CD gains) and wrist movement modalities on performance and comfort for precise distal pointing interaction. Twenty healthy participants performed a precise distant pointing task with four constant CD gains (0.6, 0.8, 0.84, and 1.0), two dynamic CD gains, and two wrist movement modalities (wrist extension and rotation) by using a mobile phone as the input device. Physiological electromyographic data, task performance, and subjective questionnaire data were collected. Comparative results show that constant CD gain is superior to dynamic CD gain and that 0.8 to 1.0 is the optimum range of values. The data showed a clear and consistent trend in performance and comfort as the CD gain increased from 0.6 to 1.0, with performance and comfort becoming progressively better, reaching an optimum at 0.84. In terms of the wrist control method, the rotation mode had smaller task completion time than the extension mode. The results of this study provide a basis for the design of remote interaction using mobile phones in an intelligent cockpit.

1. Introduction

With the widespread application of immersive large screens, AR-HUD (Augmented Reality Head-Up Display), and multi-screen linkage systems in intelligent cockpits, the human–computer interaction paradigm within cockpits is undergoing a profound revolution. As the screen layout in intelligent cockpits extends from the driver’s seat to the entire front and rear rows, the interaction space has been expanded, generating demands for both passengers and drivers to control screens without physical contact. Additionally, due to the increased interaction distance, the operation distance has extended from the traditional 30 cm for touch control to 80 cm (for the central control screen) or even more. Traditional physical knob and button interaction methods have shown obvious limitations due to the increased functional complexity. In scenarios involving large screens and remote operations, touchscreen interaction requires passengers to lean forward, which can easily lead to uncomfortable postures and affect the user experience. Distal pointing, as an interaction paradigm that bridges physical space, enables long-distance object manipulation through gestures, eye movements, or device assistance. Due to its natural and intuitive characteristics, it is gradually penetrating diverse scenarios of intelligent cockpits and becoming a key technological direction for enhancing cockpit experience (see Figure 1).

There is an extensive literature on distal pointing interactions using laser pointers, controllers, natural gestures, and everyday smart devices [1,2]. Remote interactions based on personal devices such as smartphones, smartwatches, and smart rings are now the trend in ubiquitous computing due to their ease of accessibility. Yeo et al. introduced WRIST to achieve interaction by using a smartwatch and smart ring to realize the functions of gesture and pointing [3]. Yun et al. achieved high tracking accuracy for smart TV using a smartphone and smartwatch [4]. Bi et al. designed a smartwatch-based freehand human–computer interaction system and verified the advantages of its performance by comparing it with a mouse and an air mouse [5]. The study pointed out that the performance of personal handheld devices as remote pointing devices interacting with a large screen was advantageous in terms of both task completion time and error rate [6].

Although they are simple, natural, and easy to learn and use, researchers have reported problems of fatigue [7,8] and poor accuracy [9], making distal pointing interactions a tiresome and inefficient selection method. This is because distal pointing interactions typically take place in “tabletop-less” 3D telepresence environments, where the user holds a controller or uses natural gestural control to move the cursor [10]. When the arm and wrist lack the effective support of other physical surfaces, the way the muscles move is adjusted, and slight wobbling of the user’s arm can result in reduced accuracy when interactive inputs are mapped to the display. In addition, hand muscles are more easily fatigued when gestures are performed without support. Therefore, how to improve the performance and comfort of distal pointing interaction remains an important research question to enhance the user experience [6].

The specific characteristics of the intelligent cockpit environment, such as spatial limitations and vibration interference, pose severe challenges to remote pointing technology. At present, the implementation of remote pointing relies on multiple types of technical carriers. When these technical paths are applied in the intelligent cockpit environment, there are significant bottlenecks in terms of accuracy stability, fatigue control, and scene adaptability. Pure gesture interaction without physical devices creates the problem of muscle fatigue. High-precision pointing can be achieved through wearable devices such as smartwatches and rings, but these devices have issues of accessibility and scene adaptability. At present, using personal mobile phones as remote pointing terminal devices represents a good choice in intelligent cockpits.

This paper seeks to better understand the performance and comfort of distal pointing interactions in intelligent cockpit. An experiment which compared four constant CD gains, two dynamic CD gains, and two wrist movement modalities was carried out. Arm muscle fatigue (surface electromyography of the right forearm), task performance (completion time, offset distance), subjective experience (perceived effort, perceived comfort, accuracy, ease, and fatigue) were measured. The results and their implications for the design of input in environments with full-coverage displays are analyzed and discussed. This work contributes to a new, more nuanced understanding of the performance and comfort of distant pointing interactions in large immersive spaces, and the findings can provide guidance for design in increasingly common interaction scenarios.

2. Related Works

The study notes that many factors affect the performance and comfort of distal pointing interactions, including the user-to-display distance [11], target positions [12], CD gains, and control methods [13,14], etc. This paper explores research related to the CD gain and control mode, from the perspective of the input method in human–computer interaction.

2.1. Control Display Gain

The CD gain describes the scale that maps movement in a control device to the movement in a display pointer [15]. It has been studied extensively in the context of physical and virtual control devices. By adjusting the ratio of the distance moved by the control source (input) to the magnitude of the movement of the visual elements (output), the CD gain determines the responsiveness of the interaction. Manipulating the CD gain appears to be an effective method of improving performance. There are two techniques for modifying CD gain: Constant Gain (CG) where CG is adjusted uniformly by a constant multiplier, and Dynamic Gain (DG), also known as pointer acceleration, where DG is adjusted using a nonuniform function depending on the characteristics of the pointer’s movement [16]. By applying a larger gain, small movements of the input device will be amplified in the display, resulting in increased cursor speed and improved pointing efficiency. However, problems of reduced pointing accuracy are often encountered when using a large CG.

As an interaction technique that dynamically changes the gain value according to the interaction state, DG can improve the accuracy problem by mapping the user’s controller’s movement to the pointer’s movement in a nonlinear way [16,17].

DG mainly includes two types as follows: (1) Dynamic gain based on the input speed of the controller. The principle is that when the controller pointer is far away from the target, the user will increase the speed of hand movement to quickly move the controller—using a larger gain can help the user to reach the distant target quickly and improve the interaction efficiency. When the controller pointer approaches the target, the user will slow down the movement of their hand. At this time, reducing the gain helps to improve the accuracy of the control [18]. (2) Dynamic gain based on the pointer’s position relative to the target— by increasing the gain when the pointer approaches the target, and reducing the gain around the target to improve the accuracy of the task completion [19]. However, position-based nonlinear mapping is limited in that the position of the pointing target must be known in advance to be applied in the design, and this approach is not applicable to pointing tasks where the target position is unknown.

It was found that CD gain has an effect on the precise control of smaller targets. Casiez et al. examined the impact of CD gain on mouse pointing performance on the standard desktop display [16]. They found there is a slight increase in time when selecting very small targets at high levels of CG. Pointer acceleration resulted in 3.3% faster pointing than CG and up to 5.6% faster with small targets. It is worth noting that previous research on CD gain has generally targeted controllers such as joysticks, trackballs, touchpads, drawing pads, mouse, and other controllers that need to be supported by a physical surface [15,20,21]. Even in two-dimensional interactive environments, there is a great deal of variation in the CD gain functions of these controllers and their effects on the user experience. Existing research addressing the impact of CD gain on precise control in distal pointing interactions, especially for smaller targets, is not yet sufficiently discussed [14].

2.2. Wrist-Based Interactions

Indeed, many of the distal pointing tasks involve the combination of arm and wrist motions [22]. Recent studies have shown that wrist-based interaction modes play a more advantageous role in terms of both performance and comfort. Bohan et al. reported that a small scale of hand motor movement (such as wrist movement) yielded higher operational accuracy than a large scale of hand motor movement (such as upper arm movement) [23]. Chen et al. explored the effects of postural control methods on distal pointing performance, suggesting that among the three joint-based interactive methods (wrist-, elbow-, and shoulder-based interactive methods), the wrist-based method is a better interactive method for tasks where pointing speed is the highest priority [14]. Although wrist movements are influenced by the arm’s muscle and skeletal structure, with a reasonable range of motion control, wrist interactions can trade off less effort for better reaction time [24].

Many approaches explored the idea of leveraging the wrist joint as a controller. Feiz et al. proposed wrist gestures as a modality for interacting with smartwatches [22]. WrisText proposed a one-handed text entry on a smartwatch using wrist gestures [25]. Liu proposed text entry for virtual reality using a single controller via wrist rotations [24]. The above study confirms the usability of gesture control based on wrist rotations.

However, the human wrist is very dexterous and can perform flexion, extension, radial deviation, and rotational movements [3]. Cauche et al. compared the wrist translational and rotational actuation and found that wrist translational actuation provides lower inertia and lower friction torque and is a much easier means of control compared to wrist rotational actuation [26]. The number of studies contrasting the performance of different wrist gestures in distal pointing interactions is limited. As the two main modes of wrist movement, the differences between wrist extension and rotation in terms of performance and comfort in distal pointing interactions need to be further investigated.

3. Methods

3.1. Tasks

A short-distance movement task was designed to explore precise distal pointing in an intelligent cockpit setting. The precise short-distance pointing task was designed using two types: left-right and up-down. When the left-right pointing task was performed, a 20-pixel-wide red ball and a blue square appeared on the left and right sides of the horizontal direction in the center of the screen, respectively, with a distance of 800 pixels between them, as shown in Figure 2a. Similarly, when the vertical pointing task was performed, a 20-pixel-wide red ball and a blue square appeared at the top and bottom of the vertical direction in the center of the screen, respectively, with a distance of 800 pixels between them, as shown in Figure 2b.

The red ball was the controller cursor and the blue square was the target location where the cursor moved. The subject moved the red ball to the target position where the blue square was located by controlling the sensor through wrist motion. When the red ball was moved to the target position, the blue square disappeared and simultaneously appeared on the other side of the horizontal or vertical direction, and the participant continued to move the red ball to the target position. The task was repeated 60 times in each direction. After each task was completed, there was a one-minute break before starting another task. After the completion of each stage’s tasks, there was a 5 min break before starting the next stage. Mouse-type target movement test experiments have been shown to obey Fitz’s law, which states that the time required to move an object to a target area is related to the spacing between the object and the target area as well as the size of the target area [27]. To investigate the effect of different CD gains, the size and starting position of the control blob and square were kept constant and participants always started from the same screen position.

Before starting the formal experiment, a learning stage was set up, and the participants were guided to complete the tasks using the controller. When they felt that they could complete the task without any problem, they began the stage of the formal experiment. To avoid a learning effect, the order in which each participant used a certain operation method and displayed the gain was randomly assigned. The experiment lasted for 55 min in total. Participants were told that they could stop at any time if they felt uncomfortable.

3.2. Independent Variables

The independent variables are six sets of CD gains and two wrist-based control modalities. There are four sets of constant CD gains (CG), CG1 = 0.6, CG2 = 0.8, CG3 = 0.84, and CG4 = 1. CG1 means that the speed of the pointer displayed in the interface is 0.6 times the speed of the controller movement. CG2 means that the speed of the pointer displayed in the interface is 0.8 times the speed of the controller movement. CG3 means that the speed of the pointer displayed in the interface is 0.84 times the speed of the controller movement. CG1 means that the speed of the pointer displayed in the interface is the same as the speed of the controller movement.

There are two sets of dynamic display gain (DG1, DG2). DG1(F(X) = X^0.85) represents where the speed of the display pointer increases slowly with the controller input speed (small gain rate of change), and DG2 (F(X) = 4X^0.5) represents where the speed of the display pointer increases rapidly with the controller input speed (large gain rate of change). These two functions were chosen in view of the fact that DG1 provides a smooth transition and DG2 enhances the response to high-speed movement. The relationships between different gains and the input speed of the controller are shown in Figure 3a,b, and the input speed of the controller and the output speed of the pointer under different gains are shown in Figure 3c,d. The value of the CD gains was taken with reference to existing studies, and a pilot study that included three participants. The selection was based on the relative superiority as reported verbally by the participants in the pilot study.

Two types of wrist movements were designed to control the horizontal and vertical movement of the ball. Wrist extension was used to control the movement of the ball with the palm up in the starting position. As shown in Figure 4a, wrist extension to the left was used to control the ball to the left, wrist extension to the right was used to control the ball to the right, wrist extension upwards was used to control the ball upwards, and wrist extension downwards was used to control the ball downwards.

The horizontal movement of the ball was controlled by wrist rotation with the palm starting position palm to the left, as in Figure 4b. Wrist rotation to the left controlled the ball to move to the left; wrist rotation to the right controlled the ball to move to the right; wrist rotation upward controlled the ball to move upward; and wrist extension downward controlled the ball to move downward.

3.3. Data Collection

The experimental dependent variables include three categories: physiological data, task performance data, and subjective questionnaire data. Physiological data were the surface electromyography recordings (EMG) of the participant’s right forearm, which can indicate the level of muscle fatigue. The movements involved in the experiment, such as wrist flexion, wrist extension, left wrist rotation, and right wrist rotation, are mainly dependent on the functions of the ulnar extensor carpi ulnaris and ulnar flexor carpi ulnaris muscles in the forearm muscle groups. In this experiment, these two muscle surfaces of the experimenter’s right upper limb, i.e., the ulnar extensor carpi ulnaris and the ulnar flexor carpi ulnaris, were selected as the acquisition points of the surface EMG signals, which are shown in the green part of Figure 5. The integrated electromyography (IEMG) value was calculated based on the EMG data. IEMG is the sum of the area under the curve of electromyographic signals per unit of time. It reflects the total amount of discharges of the motor units participating in the activity in the muscle within a certain period of time, i.e., under the premise of constant time, the size of its value reflects to a certain extent the number of motor units participating in the work and the size of the discharges of each motor unit.

Task performance data included the task completion time and task offset distance. Task completion time can indicate the efficiency of the task, starting from the moment the experiment was initiated and ending when the subject completed the experiment. Task offset distance can effectively reflect the accuracy of task completion by noting the deviation of the path of the ball moving relative to the horizontal and vertical targets—the larger the offset distance, the higher the task inaccuracy.

The subjective questionnaire used a seven-point Likert scale to collect participants’ subjective feelings about the task. After each task was completed, participants scored four indicators of comfort, accuracy, ease, and fatigue in completing the task according to their feelings, with comfort ranging from 0 (very uncomfortable) to 7 (very comfortable), accuracy ranging from 0 (very inaccurate) to 7 (very accurate), ease ranging from 0 (very uncomfortable) to 7 (very easy), and fatigue ranging from 0 (very unfatiguing) to 7 (very fatiguing).

3.4. Experimental Setup

In order to simulate the posture of passengers using mobile phones for remote control in the intelligent cockpit, participants were seated 1.2 m from the screen during the experiment. The height of the seat was adjusted to a comfortable position according to the height of the subject during the preparation stage, and the seat was orientated in a fixed position with armrests. Participants were asked to place their elbows on the armrests at all times during the experiment, as shown in Figure 6. Participants performed all tests on the same Lenovo-equipped laptop (Lenovo Xiaoxin 15, Lenovo Group Ltd., Beijing, China), with the task interface displayed on an external monitor with a resolution of 1920 × 1080 and a refresh rate set to 60 Hz.

The participants used a controller to complete the experimental task. To simulate the user’s hand state when using a mobile phone as a controller, the controller was designed based on the dimensions of a 6.1-inch mobile phone, with a length of 148 mm, a width of about 71.5 mm, a thickness of about 8 mm, and a weight of 172 g. The controller was equipped with a BWT901CL nine-axis Bluetooth Attitude-angle Measurement Sensor. The sensor filtered and fused the raw sensor data through the internal DMP module to obtain the solved time, acceleration, angle, and position data [28], which were transmitted to the computer via Bluetooth. The task was encapsulated in a program developed using Python 3.9.21, which received movement data from the controller at 60 Hz from the Bluetooth module, while sampling the position of the ball and the cube, and logging all the data in an excel sheet.

The Ergo Lab Smart Wearable Human Factors Physiological Recorder developed by Zinfar Technology was used to collect the EMG data of the relevant muscles of the participant in the process of accomplishing the task. The 2-channel smart wearable sensors were used for EMG data acquisition, as shown in Figure 4. To ensure the accuracy of signal acquisition, disposable bipolar surface electrodes were attached to the surface of each group of muscles in order to accurately capture the electrical signals generated by muscle activities [29].

3.5. Participants

A total of 22 participants were recruited for this experiment. Data from 2 participants were removed due to the common use of the left hand and the impossibility of precise control. Data from 20 participants were ultimately used for analysis, including 10 men and 10 women. The age range of all participants was between 18 and 30 years old (M = 22.3; SD = 1.9). They were all in good health, habitually right-handed, in good mental health, and without hearing impairment. In terms of visual acuity, they all had normal vision after correction and did not have problems such as color blindness or color weakness. Each participant was well informed about the purpose of the experiment to ensure informed consent. Each participant signed an informed consent agreement to ensure compliance and ethics of the study.

4. Results

All statistical analyses were performed with IBM SPSS 25. The Shapiro–Wilk (SW) test was used to test the normality of participants’ IEMG, task completion time, offset distance, and subjective questionnaire scores under constant CD gain (CG1\CG2\CG3\CG4) and dynamic CD gain (DG1\DG2) conditions, respectively. The results showed that the scoring data of IEMG, task completion time, offset distance suit, and all indicators of comfort, accuracy, ease, and fatigue in the subjective questionnaire followed a normal distribution. Paired-sample comparisons were performed using repeated-measures ANOVA for data following a normal distribution, and the Wilcoxon signed-rank test for paired-sample comparisons was used as a nonparametric test for data not following a normal distribution.

4.1. Constant CD Gains Comparison

For the four constant CD gains (CG1\CG2\CG3\CG4), a repeated-measures ANOVA was used to compare the differences in IEMG, task completion time, offset distance, and subjective questionnaire scoring data of the participants in the four conditions. Significant differences in CG comparison are listed in

Table 1. Significant differences in CG comparison.

Indicator	Significance (p)	Effect Size (η2)	Mean Comparison
IEMG	F = 36.903, p = 0.004	0.660	CG3 (M = 3.22, SD = 0.82) < CG4 (M = 3.67, SD = 1.13) < CG2 (M = 4.17, SD = 1.39) < CG1 (M = 5.58, SD = 2.07)
Task completion time	F = 38.330, p < 0.001	0.669	CG3 (M = 19.08, SD = 2.62) < CG4 (M = 20.89, SD = 2.93) < CG2 (M = 22.18, SD = 3.57) < CG1 (M = 29.51, SD = 7.77)
Offset distance	F = 5.037, p = 0.004	0.210	CG3 (M = 10.35, SD = 3.73) < CG4 (M = 10.61, SD = 2.75) < CG2 (M = 11.23, SD = 3.66) < CG1 (M = 12.14, SD = 4.17)
Comfort	F = 14.712, p < 0.001	0.436	CG3 (M = 5.17, SD = 0.79) > CG4 (M = 4.79, SD = 0.86) > CG2 (M = 4.86, SD = 0.64) > CG1 (M = 4.19, SD = 0.55)
Accuracy	F = 6.624, p = 0.001	0.259	CG3 (M = 5.11, SD = 0.62) > CG4 (M = 4.84, SD = 0.78) > CG2 (M = 4.71, SD = 0.58) > CG1 (M = 4.38, SD = 0.58)
Ease ratings	F = 7.882, p < 0.001	0.293	CG3 (M = 5.16, SD = 0.57) > CG4 (M = 4.95, SD = 0.73) > CG2 (M = 4.89, SD = 0.60) > CG1 (M = 4.46, SD = 0.67)
Fatigue ratings	F = 9.777, p < 0.001	0.340	CG3 (M = 2.64, SD = 1.02) < CG2 (M = 2.88, SD = 0.93) < CG4 (M = 3.21, SD = 0.99) < CG1 (M = 3.40, SD = 1.06)

.

4.1.1. Physiological Data

There was a significant difference between the mean value of the IEMG recordings of the participants in the four constant CD gain conditions (F = 36.903, p < 0.001). Further pairwise test analysis showed that there was a significant difference in IEMG values between the two pairs in the CG1, CG2, CG3, and CG4 conditions. The highest mean value of IEMG was found in the CG1 condition (M = 5.58, SD = 2.07), the mean value of IEMG was smaller in the CG2 condition compared to the value in the CG1 condition (M = 4.17, SD = 1.39), the lowest mean value of IEMG was found in the CG3 condition (M = 3.22, SD = 0.82), and the mean value of IEMG in the CG4 condition (M = 3.67, SD = 1.13) was bigger than for the CG3 condition. As the constant CD gain increased from 0.6 to 1.0, the mean value of IEMG showed a trend of gradually decreasing and then increasing, as shown in Figure 7.

4.1.2. Task Performance Data

There was a significant difference in the task completion time data for participants in the four constant CD gain conditions (F = 38.330, p < 0.001). Further pairwise test analyses showed that there was a significant difference in task completion time between the two pairs, except between CG2 and CG3, where the difference in task completion time was not significant. Among them, task completion time was the longest in the CG1 condition (M = 29.51, SD = 7.77), the second longest in the CG2 condition (M = 22.18, SD = 3.57), the shortest in the CG3 condition (M = 19.08, SD = 2.62), and increased in the CG4 condition (M = 20.89, SD = 2.93). As the constant CD gain increased from 0.6 to 1.0, the task completion time showed a trend of gradually decreasing and then slowly increasing, as shown in Figure 8.

There was a significant difference in the offset distance of the participants’ control spheres in each of the four constant CD gain conditions (F = 5.037, p = 0.004). Further pairwise test analysis showed that the offset distances in the CG1 condition were significantly different from those in the CG3 and CG4 conditions, respectively (p = 0.050, p = 0.025), and that the differences in the offset distances between the two pairs in the remaining conditions were not significant. The maximum offset distance was found in the CG1 condition (M = 12.14, SD = 4.17), followed by the second largest in the CG2 condition (M = 11.23, SD = 3.66), the smallest in the CG3 condition (M = 10.35, SD = 3.73), and a slight increase in the CG4 condition (M = 10.61, SD = 2.75). As the constant CD gain increased from 0.6 to 1.0, the offset distance to complete the task showed a trend of gradually decreasing and then slowly increasing, as shown in Figure 9.

4.1.3. Subjective Questionnaire Data

There was a significant difference in participants’ ratings of perceived comfort (F = 14.712, p < 0.001). Further paired test analysis showed that there was a significant difference in comfort scores between two pairs in the CG1, CG2, CG3 and CG4 conditions. Among them, the participants gave the lowest comfort ratings in the CG1 condition (M = 4.19, SD = 0.55), the second highest in the CG2 condition (M = 4.86, SD = 0.64), the highest in the CG3 condition (M = 5.17, SD = 0.79), and the highest in the CG4 condition (M = 5.17, SD = 0.79), and the comfort ratings given by the participants decreased (M = 4.79, SD = 0.86).

There was a significant difference in participants’ ratings of perceived accuracy (F = 6.624, p = 0.001). Further paired test analysis showed that there was a significant difference in accuracy scores between the CG1 and CG4 groups (p = 0.001). The perceived accuracy was low in the CG1 condition (M = 4.38, SD = 0.58), third highest in the CG2 condition (M = 4.71, SD = 0.58), highest in the CG3 condition (M = 5.11, SD = 0.62), and second highest in the CG4 condition (M = 4.84, SD = 0.78).

There was a significant difference in participants’ ratings of perceived ease (F = 7.882, p < 0.001). Further pairwise test analysis revealed significant differences in perceived ease between CG1 and CG2, and between the CG1 and CG4 pairs. The perceived ease rating was highest in the CG3 condition (M = 5.16, SD = 0.57) and lowest in the CG1 condition (M = 4.46, SD = 0.67); the perceived ease ratings were higher in both the CG2 condition (M = 4.89, SD = 0.60) and the CG4 condition (M = 4.95, SD = 0.73) than in the CG1 condition.

There was a significant difference in participants’ ratings of perceived fatigue (F = 9.777, p < 0.001). Further pairwise test analyses revealed significant differences in fatigue ratings between CG4 and CG1, and between the CG4 and CG3 pairs (p = 0.003, p < 0.001). The highest fatigue ratings were found in the CG1 condition (M = 3.40, SD = 1.06), the second highest in the CG4 condition (M = 3.21, SD = 0.99), the third highest in the CG2 condition (M = 2.88, SD = 0.93), and the lowest in the CG3 condition (M = 2.64, SD = 1.02).

As shown in Figure 10, as the constant CD gain increases from 0.6 to 1.0, the perceived comfort, accuracy, ease and fatigue showed a gradual trend of getting better, with 0.84 being the best, and then getting worse.

4.2. Dynamic CD Gain Comparison

For the two dynamic display gains (DG1 and DG2), a repeated-measures ANOVA was used to compare the IEMG, task completion time, offset distance, and subjective questionnaire scores under the two conditions. The results showed that there were no significant differences in IEMG and task completion time between the two groups of dynamic CD gains, and there were no significant differences in participants’ comfort scores, ease scores, and fatigue scores in the subjective questionnaire.

There was a significant difference in the offset distance of the blob movement between the two groups of dynamic CD gains (F = 106.407, p < 0.001). The offset distance was smaller in the DG1 condition (M = 11.27, SD = 3.11) and larger in the DG2 condition (M = 13.73, SD = 2.74).

There was a significant difference in participants’ accuracy scores (F = 13.514, p = 0.002). The accuracy scores were higher in the DG1 condition (M = 4.55, SD = 0.73) and lower in the DG2 condition (M = 4.12, SD = 0.77). Significant differences in the DG comparison are listed in

Table 2. Significant differences in DG comparison.

Indicator	Significance (p)	Effect Size (η2)	Mean Comparison
Offset distance	F = 106.407, p < 0.001	0.849	DG1 (M = 11.27, SD = 3.11) < DG2 (M = 13.73, SD = 2.74)
Accuracy ratings	F = 13.514, p = 0.002	0.416	DG1 (M = 4.55, SD = 0.73) > DG2 (M = 4.12, SD = 0.77)

.

4.3. Constant and Dynamic CD Gain Comparison

For the constant and dynamic CD gains, participants’ IEMG, task completion time, offset distance, and subjective questionnaire scores were compared between the two conditions using repeated-measures ANOVA. The results showed that there was no significant difference between the IEMG of the participants in the two conditions. There was a significant difference in task completion time (F = 5.143, p = 0.035). The task completion time was shorter in the constant CD gain condition (M = 22.91, SD = 3.73) and longer in the dynamic CD gain condition (M = 24.05, SD = 2.75). There was a significant difference in offset distance (F = 29.009, p < 0.001). The offset distance was smaller in the constant CD gain condition (M = 11.16, SD = 3.45) and larger in the dynamic CD gain condition (M = 13.73, SD = 2.74).

There was a significant difference in participants’ comfort scores on the subjective questionnaire (F = 13.146, p = 0.002). The comfort scores were higher in the constant CD gain condition (M = 4.75, SD = 0.59) and lower in the dynamic CD gain condition (M = 4.29, SD = 0.57). There was a significant difference in accuracy scores (F = 9.538, p = 0.006). The accuracy was higher in the constant CD gain condition (M = 4.76, SD = 0.46) and lower in the dynamic CD gain condition (M = 4.33, SD = 0.70). There was a significant difference in ease scores (F = 13.840, p = 0.001). Ease was higher in the constant CD gain condition (M = 4.86, SD = 0.52) and lower in the dynamic CD gain condition (M = 4.43, SD = 0.63). There was a significant difference in fatigue scores (F = 19.453, p < 0.001). In particular, fatigue scores were higher in the dynamic CD gain condition (M = 3.77, SD = 0.94) and lower in the constant CD gain condition (M = 3.03, SD = 0.91). Significant differences in the CG and DG comparison are listed in

Table 3. Significant differences in CG and DG comparison.

Indicator	Significance (p)	Effect Size (η2)	Mean Comparison
Task completion time	F = 5.143, p = 0.035	0.213	CG (M = 22.91, SD = 3.73) < DG (M = 24.05, SD = 2.75)
Offset distance	F = 29.009, p < 0.001	0.604	CG (M = 11.16, SD = 3.45) < DG (M = 13.73, SD = 2.74)
Comfort ratings	F = 13.146, p = 0.002	0.409	CG (M = 4.75, SD = 0.59) > DG (M = 4.29, SD = 0.57)
Accuracy ratings	F = 9.538, p = 0.006	0.334	CG (M = 4.76, SD = 0.46) > DG (M = 4.33, SD = 0.70)
Ease ratings	F = 13.840, p = 0.001	0.422	CG (M = 4.86, SD = 0.52) > DG (M = 4.43, SD = 0.63)
Fatigue ratings	F = 19.453, p < 0.001	0.506	CG (M = 3.77, SD = 0.94) < DG (M = 3.03, SD = 0.91)

.

4.4. Wrist Extension and Rotation Comparison

The IEMG, offset distance, comfort score, accuracy score, ease score, and fatigue score data of the participants in the rotated interaction mode condition did not follow a normal distribution. A paired-samples test using the Wilcoxon rank sum test found no significant differences in participants’ IEMG, offset distance, comfort scores, accuracy scores, ease scores, and fatigue scores between the two interaction style conditions, with a significant difference found only in task completion time (Z = 2.448, p = 0.014). Comparison of the medians revealed shorter completion times when using the rotational approach to maneuvering (M = 21.025, SD = 2.279) and longer completion times when using the translational approach to maneuvering (M = 22.125, SD = 1.893).

5. Discussion

In this study, we compared differences in participants’ IEMG (degree of fatigue of the forearm muscles), task completion time and offset distance (task performance), and subjective feelings (questionnaire) between conditions of different sizes, types of CD gains, and wrist control modes through an experiment in which a short-distance pointing task was completed in a distal setting.

When using different sizes of CGs (CG1 = 0.6, CG2 = 0.8, CG3 = 0.84, CG4 = 1), participants showed significant differences in IEMG, task completion time, and offset distance, and compared to the other CGs, participants in the CG3 condition had smaller task completion times, offset distances, and IEMG, and subjectively presented significantly higher task completion comfort, accuracy, ease, and lower fatigue. It suggests that CG3 has the highest task completion performance and accuracy, which effectively reduces muscle fatigue and enhances subjective comfort. The results of this experiment differ from the suggested CD gain range of existing studies. Existing studies have suggested a CD gain range greater than 1.0 for the wrist-based interaction method, whereas this experiment found 0.84 to be more advantageous than 1.0 [12]. The reason for such differential results might be that the experimental tasks in this study were more refined. When the subject was at a distance of 1.2 m from the screen and only moved 800 pixels from the desktop display, this was equivalent to a very precise position of hand movement. Therefore, a display gain of less than 1 was required to avoid exceeding the range. In addition, participants’ IEMG, time to complete the task, offset distance, and subjective perceptions of comfort, accuracy, and ease gradually decreased as the CG value approached 1.0 from 0.6, optimized at 0.84, and then slowly increased. The consistent trends of the different types of data corroborate each other. It indicates that the optimal value of constant CD gain ranges from 0.8 to 1 when accomplishing a short-distance pointing task in distal pointing settings.

There was a significant difference in the task offset distance when using different sizes of DGs (DG1, DG2). No difference was found in the IEMG data and task completion time. DG1 was effective in reducing the offset distance during the task compared to DG2 and presented a significantly higher subjective accuracy of task completion than DG2. This suggests that dynamic CD gains that increase gently with controller motion speed are more beneficial for task performance and interaction usability when accomplishing short distal pointing tasks in distal pointing settings. Given the significant influence of the design of the dynamic gain function on the conclusion, different parameters of the dynamic display gain may lead to performance differences. Future research will need to focus on comparative experiments of the gain function to test the effects of functions with different parameters and types.

There were no significant differences in the IEMG when using different types of CD gains (CG, DG), and there were significant differences in task completion time and offset distance. Compared to DG, CG was effective in reducing task completion time and control offset distance, and presented significantly higher subjective comfort, accuracy, ease, and lower fatigue than dynamic gains, significantly improving task performance and interaction usability. This suggests that constant CD gain significantly improves task performance and interaction usability when accomplishing short-distance pointing tasks in distal pointing settings. This is in contrast to the findings for the mouse controllers in 2D interfaces, where the dynamic display gain performance and experience is superior.

Significant differences were found only in task completion times when using different types of wrist movement modalities (wrist extension and rotation). Completion times were slightly shorter when using the wrist rotation maneuver than when using the wrist extension maneuver. However, no significant differences in other indicators, such as accuracy or fatigue, were reported. Therefore, the advantageous role of wrist rotation mode in distal pointing settings awaits further experimental verification.

6. Conclusions

This study explored the effects of different control display gains and wrist movement modalities on performance and comfort in a precise distal pointing task in the setting of intelligent cockpits. Differences in participants’ IEMG (degree of fatigue of the forearm muscles), task completion time and offset distance (task performance), and subjective feelings (questionnaire) were compared between conditions through an experiment in which a short-distance pointing task was completed. The results suggested that the constant CD gain was better than the dynamic CD gain, with an optimum value between 0.8 and 1.0. And the wrist rotation mode is better than the wrist extension mode in the task completion time. The results of this study inform the design of the input method for remote interactions that use mobile phones in intelligent cockpits.