Programming IoT-spaces: A User-Survey on Home Automation Rules

Danny Soares, João Pedro Dias, André Restivo and Hugo Sereno Ferreira

ABSTRACT. The Internet-of-Things (IoT) has transformed everyday manual tasks into digital and automatable ones, giving way to the birth of several end-user development solutions that attempt to ease the task of configuring and automating IoT systems without requiring prior technical knowledge. While some studies reflect on the automation rules that end-users choose to program into their spaces, they are limited by the number of devices and possible rules that the tool under study supports. There is a lack of systematic research on (1)~the automation rules that users wish to configure on their homes, (2)~the different ways users state their intents, and (3)~the complexity of the rules themselves --- without the limitations imposed by specific IoT devices systems and end-user development tools. This paper surveyed twenty participants about home automation rules given a standard house model and device's list, without limiting their creativity and resulting automation complexity. We analyzed and systematized the collected 177 scenarios into seven different interaction categories, representing the most common smart home interactions.

Presented at IoTSS, part of ICCS 2021: https://www.iccs-meeting.org/iccs2021/

Transcript

1. Programming IoT-spaces: A User-Survey on Home Automation Rules Danny Soares, João Pedro Dias, André Restivo and Hugo Sereno Ferreira {up201505509,jpmdias,arestivo,hugosf}@fe.up.pt International Workshop on Computational Science in IoT and Smart Systems (part of ICCS 2021)
2. Index ▪ Context ▪ Related Work ▪ Survey Process ▪ Results & Categorization ▪ Analysis ▪ Final Remarks
3. Context ▪ Smart homes are a primary example of an IoT-space. ▪ Commonly, the smart home users do not have technical background. ▪ The complexity of IoT systems (e.g., heterogeneity and scale) makes it hard for end-users to configure and automate them. ▪ This led to the (re)birth of several low-code programming strategies for end- user development (trigger-action programming, visual programming, etc.) ▪ It becomes of paramount importance to understand what the end-users wish to automate, how they state their intents, and grasp into the users’ programming mental models.
4. Related Work (1/2) ▪ Dey et al., circa 2006, gathered 371 automation scenario descriptions from 20 participants. Almost all the participants (95%) stated their automation rules in if-then fashion, and around 23.5% of the rules used explicit Boolean logic. ▪ Brush et al., circa 2010, conducted a smart home in-situ study concluding that most user interactions were either direct actions or rule-based. ▪ Ur et al. studied the use of trigger-action programming in smart homes (using the IFTTT service), finding out that only 0,8% of the total rules analyzed use physical devices as triggers and 1.3% use physical devices as actions. Mi et al. carried a similar study concluding that “the fact that most tasks (in the smart home context) we want to automate are indeed simple”. ▪ Ammari et al. study on the use of voice assistants in smart homes, finding out that only five study participants (out of 170) defined automation rules.
5. Related Work (2/2) ▪ Most of the works that study automation rules in smart home context are too open and vague in terms of the diversity and complexity of user-defined rules. ▪ Studies that focus on specific tools (e.g., IFTTT) are limited by the features of the tool itself, not encompassing all the automations that a user could have. ▪ Some studies point that the limitations posed by off-the-shelf IoT devices are the definer of the created rules (limiting their complexity). ▪ We agree that the limitations posed by the devices and the current IoT ecosystem limit what end-users can program them to do and that most of the rules are indeed simple by nature, as the number of inhabitants and devices increases, the resulting operational context can be complex to model and reason about.
6. Survey Setup (1/2) ▪ Home model: a garage, a front patio, a pool, a garden, a living room, a kitchen, one bedroom, and a bathroom.
7. Survey Setup (2/2) ▪ Device List (suggestions): ▪ Motion, temperature, humidity, smoke, and air quality sensors; ▪ Security cameras; ▪ Controllable lights; ▪ Controllable windows and blinds; ▪ A/C system; ▪ Robot vacuum cleaner; ▪ Sound system; ▪ …
8. Methodology, Population and Collected Data ▪ List of suggested devices and home model provided; ▪ Google Forms with an open question for rule writing (as many as they whish); ▪ 20 participants from different educational fields and ages. ▪ 177 automation scenarios collected (dataset available).
9. High-level Categorization ▪ Sensors and actuators ▪ Scenarios that only use sensors and actuators, where the sensors trigger the actuators; ▪ Actuators on schedule ▪ Scenarios where actuators are triggered on a fixed schedule; ▪ Actuators on time interval, with sensors ▪ Scenarios that combine sensors information and time intervals to trigger actuators; ▪ Sensors with timers ▪ Scenarios that combine sensors information and time intervals to trigger actuators; ▪ Actuators with timers ▪ Scenarios where the actuators are triggered when the status of the sensor does not change for some time; ▪ External services ▪ Scenarios that depend on external services to trigger the actuators. ▪ One-time actions ▪ Scenarios that are meant to happen only once, instead of being recurring
10. Analysis Scenarios per Category and Room Mention
11. Analysis Resulting Ecosystem
12. Key Takeaways ▪ Some of the submitted rules would need to be “rewritten” in a programmable fashion; ▪ The most common way of specifying scenarios is by using the structure “when condition, then action” or “action when condition” (TAP-like); ▪ Users also tend to specify similar (or equal) scenarios using different expressions, granularity and forms; ▪ We consider this as a mostly direct result of the different backgrounds. ▪ Approx. 28% of all the submitted scenarios mention “turn on” actions; ▪ Integration with external services is only directly mentioned once (but there are several indirect mentions); ▪ Some rules are too generic, e.g., “Shut down all unnecessary devices”, which would require that the IoT system had some degree of contextual awareness to be able to execute them; ▪ Some rules also depend on a priori defined preferences (e.g., set to a predefined temperature).
13. Final Remarks ▪ There are some identified threats to the validity of this study, which include: ▪ Comprehension of the submitted scenarios, sample size, level of expertise of participants and the observable little variety on the scenarios per category. ▪ The most common way to define automation scenarios is by conditional programming, like the approach used by applications such IFTTT. ▪ While most of the scenarios could be easily mapped into TAP rules, the rules that do not follow such model appear as a challenge which is mostly ignored by existing solutions. ▪ Contextual awareness (e.g., knowing user preferences) adds value to low-coding solutions. ▪ The dataset is available for further analysis. We believe that a formalization of the categorization process as well as expanding the dataset would improve the study conclusions and provide even more future research directions.
14. Thank you! Dataset: https://doi.org/10.5281/zenodo.4531395 Programming IoT-spaces: A User-Survey on Home Automation Rules Danny Soares, João Pedro Dias, André Restivo and Hugo Sereno Ferreira {up201505509,jpmdias,arestivo,hugosf}@fe.up.pt International Workshop on Computational Science in IoT and Smart Systems (part of ICCS 2021)