HugBot: Designing a Persuasive Robot for Smartphone Addiction Control

The convenience of mobile applications and social networking services in our daily lives has caused an increase in addiction to mobile devices, posing many adverse effects. While there are many products on the market that combat this addiction using coercion, there have been few designed using the conscious abilities of persuasion. In this paper, we develop a persuasive robot, 'HugBot,' to help control smartphone overuse. HugBot responds accordingly in verbal and non-verbal cues to the user's actions, discouraging cellphone use. Furthermore, we hope to discover new paradigms for using stuffed-toy-based persuasive robot systems in user-centric addiction control.


INTRODUCTION
Smartphones have become a ubiquitous part of society, with 6.4 billion users worldwide [4].These devices, with uses in communication, entertainment, productivity, and many more, have undeniably enhanced convenience and enjoyment in our lives.However, the presence of smartphones has also created challenges, notably the struggle for users to maintain a reasonable usage time.A general consensus among researchers acknowledges the existence of cellphone addiction, with some even arguing that it might be the greatest addiction of the century [12] .Studies suggest that 22.1% of adolescents and 27.9% of young people are considered to be cellphone addicts [1].
Extensive research has been done on the efects of such addiction for over a decade [3].Overuse has been linked to poor sleep time and quality, as evidenced by Sahin et al's fnding that problematic mobile phone usage correlates with a lower Pittsburgh Sleep Quality Index [8] .Moreover, cellphone addiction has also been linked to substance abuse, with Sanchez-Martinez and Otero suggesting a strong relationship between cellphone abuse, school failure, smoking, and consumption of marijuana and other drugs [15] .
Given the prevalence of this issue, numerous digital and physical products have been manufactured with the aim of reducing the user's screen time.A popular solution is the "lock box, " where the user places their device inside a box, sets a timer, and the box is locked until an emergency button is pressed or the set time has passed.This approach takes away the conscious decision of using a phone while the device is in the box, making it a coercive solution.
In contrast to other felds such as public health [11], policy [13], and religion [9] where the diference in efectiveness of coercive versus persuasive approaches has been extensively studied, the same distinction in the context of smartphone addiction still remains to be explored.The market heavily leans towards coercive solutions, with limited emphasis on persuasive methods that empowers user with the ability to make a conscious decision with their phone [14].Recognizing this gap, we decided to design a system that increases self-control over smartphone usage in a persuasive lens.
Simultaneously, insights from human-robot interaction research suggests that social robots are suited to infuence human behaviour or thinking [5].Numerous studies has proved that robots are capable of persuading humans in behavioral change [2], with applications in weight loss [6], energy conservation [10], and health self-management [7].For such reason, hypothesizing that this persuasive capability in robots would help increase self-control over smartphone usage, we designed a personal robot with an interface of a stufed toy to help decrease screen time to combat smartphone addiction.

DESIGN METHODOLOGY 2.1 Robot Architecture
While many types of robots have anthropomorphic features that make them ideal candidates for persuasion, we chose a stufed animal interface for their familiarity [16] and efectiveness in bonding [8].
To keep the robot's production cost reasonable for future widespread use, we use a store-bought stufed animal in which we insert a rigid 'skeleton' inside its arms and head.The ends of the skeleton are all attached to a series of three mini servo motors, each controlling the hugging motion of the arms or the nodding motion of the head.The stufed animal sits atop a base with a motion sensor facing the user that can sense when the user approaches the robot.On the base, we also include a speaker that provides verbal feedback on the user's actions and a button-powered contraption that senses smartphone placement.All components are connected to an Arduino Uno microcontroller that is responsible for controlling the motors, sensors, and speaker.Our robot is named 'HugBot' for its signature feature of 'hugging' the user's cellphone.

Use Setting
The size of the HugBot allows it to comfortably sit on one's desk.At the initial stage, we are expecting HugBot to be utilized in personal spaces, such as a private ofce or home.The implementation of HugBot in a more open context like corporate ofce or public spaces will introduce a new kind of interaction between the user and nonuser.For the sake of simplicity in analyzing the efectiveness of HugBot in isolation, we will focus on the simplest frame with only the robot and user in play.Our aim is for the HugBot to become a benefcial intervention in those situations.

Action Algorithms
The HugBot's actions fall into three patterns, which is pictured in Figure 1, 2 and 3.In the scene pictured in Figure 1, the HugBot starts in its initial state where its arms are extended and its head is up.This creates a space for the user to place their smartphone.Once the user sets their device in the slot, the button below the slot will be activated, and HugBot will close its arms and head around the device, activating its "Hug State." The arms and head will provide just enough grip so the device will not fall over, but weak enough so the user can easily remove it.In this state, when the user reaches out for their smartphone, a motion sensor attached to HugBot's base will detect the user's hand.This detection triggers the HugBot to display discomfort with the aim of discouraging the user from further reaching out for their device.Specifcally, it will make a frustrated noise through a speaker attached to the base while tightening the grip of its arms.At this point, the user is left with two choices: proceeding to pick up the phone or opting not to take the phone.
If the user is not persuaded to choose the former, the situation will move to Figure 2. In this scene, as the device is taken away, the button below HugBot will no longer be pressed.This triggers HugBot to display sadness by playing a sad noise.The HugBot will then go back to its initial state with its arms wide open and head facing up.If the user is persuaded to choose the latter, the situation will move to Figure 3.In this scene, as the user's hand moves out of range of the motion sensor, HugBot will display happiness by making a happy noise, maintaining its Hug State.At the current state of development, HugBot will maintain its Hug State indefnitely until the device is taken away and the button below HugBot is released.After assembling the frst prototype of our robot, we began our initial testing of HugBot, in which we address the following key questions

Initial Short-Term Testing
• What are your opinions on the interface?
• Does the stufed animal persuade you to gain more control over smartphone usage?• What improvements could be made to improve the usability of this robot?To answer these questions, we collected 10 random participants, predominantly college students.Our demonstration included the three scenes explained in 2.2 Action Algorithms, as well as a free interaction session where we allowed users to touch the form of the robot.
Firstly, the responses on the interface were mainly positive, with users expressing appreciation for the aesthetically pleasing and cute design.However, we noted negative feedback about the open electronic components of the robots, highlighting an area for improvement in our next iteration.
Secondly, we sought to understand if HugBot efectively infuenced users to exert more control over smartphone usage.Many participants acknowledged the emotional infuence of taking away the smartphone from the robot, making the HugBot efective as a persuasive robot.Nevertheless, a user raised concerns about the unfamiliarity of an non-locked system, expressing uncertainty about its efcient use.
Lastly, participants were encouraged to provide insights into potential enhancements for improving the HugBot.As mentioned in the frst question, there was a recurring theme centered around the visibility of wires and breadboards in our prototype.
In summary, our initial testing validated our hypothesis that a persuasive robot such as HugBot could positively infuence smartphone usage, leading to a more long-term quantitative study.Negative feedback served as valuable input of avenues on improvement for the development of HugBot.

Further Testing
Given the overall positive fndings from our initial short-term testing, our next logical step would be to conduct a more long-term study where we test the efectiveness of the stufed animal interface.This extended investigation will be conducted over a week, to gather more quantitative data as well as view the long-term efects of such solutions.
Participants will be randomly assigned to three groups.The frst group will be asked to place their cellphone inside a transparent box without locks, the second group will be asked to place their cell phone on HugBot's dedicated slot, and the third group, serving as a control, will not be given instruction on smartphone placement.Despite having unrestricted access to their devices, participants in all groups will undergo an educational session outlining the adverse efects of excessive smartphone usage.They will be prompted to use their device as little as possible in the testing period.
To study the impact of each interface, we intend to analyze the participant's average screen time not only during the experiment but also in a random week before and after the testing period.This dual time-frame approach will allow us to capture immediate efects as well as sustained changes in usage behavior, providing a holistic view of the efectiveness of the interfaces.
The anticipated outcomes will ofer insights into the capabilities of the HugBot as well as test the efectiveness of persuasive mechanisms for smartphone usage.

Future Improvements
Feedback collected on the design of HugBot during our initial testing highlighted three main areas for improvement.
First, HugBot's interface needs additional refnement.Hiding electronic details from the user is a priority, as it will allow the user to believe it is interacting with a stufed animal rather than a robot.A proposed solution involves installing a base at which HugBot will sit on, masking all the electronic components.Moreover, to make the robot more lifelike, features such as reactive touch, randomized movement, and responses to voice input are being considered to bridge this human-robot gap.
Additionally, HugBot primarily attempts to persuade users using audio information.While audio is a powerful tool to convey emotion, we wish to add more non-verbal cues to further persuade users to make the conscious decision of not using their phone.For example, further action generation, such as a closer grasp of the device or head twitches, may be efective at making the emotions more believable.
Lastly, HugBot is limited to only two scenarios at the moment: the user either taking or not taking the device.These two scenarios will cover the minimum amount of situations, but we believe that there are many more possible scenarios that should be considered in order to make the interaction between the robot and user comprehensive and authentic.For instance, HugBot may display happiness when the user does not reach out to their device for a long period of time.Also, if the user sets down their device repeatedly, HugBot may communicate encouragement and support to the user.Finally, there could be diferent interaction scenarios based on other factors in the room, such as other users or other similar robots.These proposed enhancements may improve the HugBot's capability as a persuasive robot.

Conclusion
Our design of the HugBot was met with mostly positive feedback and was perceived as persuasive, providing a solid basis for longterm research can be done to test 'coercion vs persuasion' in increasing control over smartphone usage.Moreover, if proved efective by such, the conceptual interface developed in this study could be used for user-controlled action intervention in a broader variety of contexts, hopefully bringing about a paradigm where more personal robots support our everyday actions.

Figure 1 :
Figure 1: Smartphone is given to HugBot

Figure 2 :Figure 3 :
Figure 2: User reaches for smartphone and decides to take it

Figure 4 :
Figure 4: HugBot in its initial state and 'hug' state