Teaching Tangible Interaction: Beyond the Kit of Parts
James Hallam, Georgia Institute of Technology
Tangible Interaction Design has grown into a core part of Industrial Design and Interaction academic programs in recent years, largely following the wide release of the Arduino platform. Integrating electronics knowledge into a form-making design curriculum provides many new opportunities for students, but comes with a set of unique challenges — especially for a student body without engineering backgrounds. The technology itself can be daunting, but it can also dominate the design process, as students grapple with learning to use the new tech.
The Interactive Product Design Lab (IPDL) at the School of Industrial Design at Georgia Tech has been working to update and expand our curriculum to better support our students — who are now highly invested in learning about Tangible Interaction Design. Our introductory course now serves both undergraduates and graduates, both with design backgrounds and without, in the same interdisciplinary learning environment.
As the course has expanded in scope, however, it has raised some new questions about what the nature of an “Introduction to Tangible Interaction Design” course is supposed to be. The topic is already a hybrid of Industrial Design and Interaction Design, with elements of HCI, Physical Computing and many other neighboring disciplines included. The question then is whether there is some unique curriculum at this multidisciplinary intersection — some way of making and teaching that defines “tangible interaction design”, or whether it is a diluted introduction to each of the constituent disciplines.
After teaching a course on this topic for five years, I have tried to answer this question by refining our curriculum for this course, with the goal of helping students manage the complexity of the subject matter. This has led to the introduction of a new integrated design process, the addition of an “action” model to existing form and function models, and the introduction of sensor creation through material exploration. I believe these additions will help out students better manage these separate ways of thinking and will demonstrate a native way of making tangible interactive products.
Background
Originated by Prof. Jim Budd, the Introduction to Interactive Products course has been taught in one form or another at 4 different schools (University of Illinois Urbana-Champaign, Carleton University, Emily Carr University of Art + Design, and Georgia Tech) since the mid-90s. Conceived as a way to introduce Industrial Design students to Interactive Technology, the scope of the course grew with each iteration, through Jim’s founding of the IPDL in 2011.
After a few years as a teaching assistant in the lab, I agreed to start teaching the course in 2014 — a period of time that came with increased interest in the course from students outside of the ID program. This new interest came with a technological shift with the students — a decade earlier, it was unlikely that most industrial designers would have felt comfortable with rudimentary circuit design. Now, students entering the course now are highly likely to have used a microcontroller platform (like Arduino) prior to starting the course — even the ID students. It seemed no longer sufficient to introduce them to the technology, but was now necessary to integrate that learning with interaction design and product design fundamentals.
Design students now have elevated expectations for technical complexity and seem eager to set aside some basics of design and form-making in favor of something that is functional at an early stage. This seems to stem from a belief that function is more valuable — a belief that is emphasized by the focus on Arduino kits and breadboard sensor circuits. This leads them to grapple with increasingly complex problems of integration, as inevitably they try late in the design process to force their working models into a finished form model, capable of being tested.
Design Process Precedents
To deal with this, we updated the design process model we teach in the class and structured the course to guide the students through (at least) three rounds of iterations. This approach is based on a hybrid of two well-understood design process models, and the introduction of three distinct types of model-making.
The process model starts with an adaptation of Damien Newman’s Design Squiggle. This is not a model that a team can build a detailed project plan around, but it shows the most essential path of progress for a design team — moving from uncertainty of what to build, to clarity of what to build. To guide designers through this journey of uncertainty, we encourage them to make models — artifacts whose creations helps the designer to understand whether they are building the right thing.
These models have a role to play at the beginning of the process, at the point of greatest uncertainty. We highlight the prototype as a special kind of model — one that is prototypical of the final design — and the last model that designers will typically have control over the before the production process begins. At the end of the squiggle is the design itself — the resolved pattern that describes the product that will be manufactured and replicated. As with most projects that feature objects requiring tooling and fabrication, the cost of making changes is much lower during the earlier modelling stage, than the later design stage — making it vital to get feedback on the fit of the project solution earlier than later.
The second influential design model is the Double Diamond — popularized by the UK Design Council. This model emphasizes iteration, through periods of divergent thinking (taking in new information) and convergent thinking (refining and eliminating information). This process creates feedback loops and inflection points, which measure the success of the design effort against the project brief. An evidence-based approach converts the diamonds into tests of a hypothesis about the fit of a given project solution. As with the squiggle, designers move through this process by making models, and then evaluating the success of what they have built.
Action Model
The models built during these processes have typically followed the traditional pattern of form and function models. These may include sketching, foam models, and appearance models on the form side, and schematics, breadboard circuits, and printed circuit boards on the function side. These approaches are valuable, but they don’t directly address the design of the interaction itself, which is vital to the success of the project. Here, we saw the opportunity of introducing an “action model” that incorporates Ix prototyping techniques — such as bodystorming, paper prototypes, and storyboards. The goal of the action model is to directly design and test the interaction and use it as part of the feedback loop to match the form and function.
This produces a set of “looks-like”, “works-like”, and “acts-like” models that are each relatively easy to develop and refine using their own techniques, but that collectively inform the overall design. It is easier to keep the modelling processes separate — to allow for rapid iteration at a lower cost — but to use the information gained from each new model to influence the next. Eventually, as the designer starts to understand the relationship of all the design elements, they work over the project to integrate these separate modeling processes into a single integrated prototype.
Integrated Modelling Process
The integrated modelling process is therefore the adaptation of the Design Squiggle and Double Diamond design processes to take advantage of the three distinct modelling processes, and to integrate them into a single system. This process model starts with each of form, function, and action models separated, with the goal of eventually integrating them. To do this, the designer starts with one, develops it, and creates the next model type with information gained from the first. This continues though a period of divergent thought until the designer feels they have enough information to test their design hypothesis, at which point they move to converge and integrate the three models together.
The resulting integrated model allows them to measure their success against the requirements for the project, and to make plans for the next stage of the design process. After evaluating the integrated model, they return to the separate modelling streams, with the intent of integrating them again when they have gathered enough new information. Returning to the separate modelling processes allows for more rapid iteration and exploration, while converging to an integrated model encourages repeated testing. In this way, the designer begins to move through the uncertainty of the design process and converges on the end goal of the resolved prototype.
Sensors
As we started to implement this process in the course, we saw some challenges in using the typical Arduino kits (dev board, sensor breakouts, connectors, etc..) as they imposed many formal constraints on the student’s projects. The breakouts are wonderful for guiding students to understand the sensor’s function on a breadboard but are challenging to integrate into the student’s modelling materials (cardboard, basswood, paper, fabric) in the early stages of their design development.
It is here that we were inspired by Hannah Perner-Wilson’s work (and her collaborations with Leah Buechley and Mika Satomi) that led to the kit-of-no-parts paper in 2011. As documented on Hannah’s “How To Get What You Want” site, she has more than a decade’s worth of development of craft-based e-textile and DIY sensing and computing creations. The kit-of-no-parts is her proposal that this approach — which replaces a kit of finished parts with a set of material exploration techniques and recipes — leads to projects models that are more expressive and understandable.
Building on this, we created a new project that began with a sensor creation workshop that replaced the kit parts we had previously used. Making sensors has not been typically thought of us as part of an introductory class, as it is thought to be too technical. One of Hannah’s great contributions was showing the broad range of sensors that could be built out of common craft materials, using techniques borrowed from papercraft, knitting, and sewing. We determined that by introducing our class to a basic analog pressure sensor, contact switch, and capacitive sensor, that there was a whole library of options that were now within reach.
Sensors, at their most basic, are created through the combination of material and pattern-making, both of which are inherently native forms of exploration for designers. By integrating the sensors directly into their projects, rather than creating a void space to insert a pre-fabricated sensor package, our students were able to able to explore interactions that were deeply integrated with the artifacts they were creating. Through this work, we saw the class take a major leap forward in their creativity and in the freedom they had to explore form, function, and action.
Results
We have now run the project for many semesters and have seen the results in the student’s work. Their final project asks to create a video game that is playable with a custom controller that allows for a novel interaction between the user and the game. This controller must work entirely with sensors the students develop and integrate themselves and must be capable of playing a game they develop in Processing. After three rounds of iteration, guided by the Integrated Model Design Process, the student present their functional games for review, and display them as part of the Lab’s end of year exhibition.
I now believe these approaches represent a native way of making with tangible interaction design. Through the introduction of the modified design process, the ‘acts-like’ model, and the sensor creation workshop, we have seen a big shift in the student’s engagement, understanding, and delivery of this final project. They are no longer bound by the formal requirements of the kit sensors, and free to explore many novel types of interaction.
We are still refining the curriculum with feedback from our students, and I would be happy to hear from anyone who has tried to implement a course with similar requirements. For any else who is interested in teaching a similar course, I would recommend considering the following:
- Use Acts-like models to explore interactions independent of form/function
- Use integrated models to test design theories, and to seed exploration with new insight
- Use material exploration to make custom sensors, and explore new interactions
Related work by the author
Exploring the Next Wave
Jim Budd, Kevin Shankwiler & James Hallam (2014). INNOVATION, Vol. 33 №4, p. 26: https://issuu.com/idsa.innovation/docs/innovation_winter14_/28
Video of the Presentation
About James
Following 20 years as an interface designer, strategist, and design consultant, James is now pursuing his PhD in Industrial Design at Georgia Tech. His research is focused on wearable technology and stroke rehabilitation, looking at the impact design choices have on patient adherence. James works as an Instructor at Georgia Tech — teaching “Introduction to Interactive Products” — which covers interactive prototyping, technical skills, design process, and storytelling. He is pursuing a Minor in Technology Commercialization through the GT School of Business, and has spent the last 5 years consulting as a Design Strategist. James grew up on a small island on Canada’s west coast, which makes him an expert in Gore-Tex, sandstone tide pools, and adding maple syrup to everything.
Specialties: Wearable Technology, Industrial Design, Design Strategy, Service Design, Interaction Design, Design Research, Sustainable Design, Communication Design.
