While there are SDKs that can detect objects/images without structured patterns, these are more complex to use (requiring a fairly high degree of technical proficiency), less reliable (especially at longer ranges, with low-detail references, or in poor lighting conditions) and must be pre-registered with the reader device (unlike a standards-based scheme e.g., QRCodes). As a consequence, they are rarely used beyond special-purpose experiences requiring a custom app. Importantly, in both cases, the data payload of printed images is inherently static (i.e., one marker maps to one value). To achieve multiple values or states, lay users often use multiple markers on the same document. 

In contrast to printed markers, electronic tags are dynamic, able to change in value and even relay changes in their state or environment. This sophistication increases cost and complexity — either the tag itself is reasonably capable (e.g., microcontrollers with wireless communication),or low-cost but require a specialized reader (e.g., RFIDtags). In both cases, the price and skill required to deploy electronic tags is much higher than printed fiducial markers, and for this reason, it is rare to see such systems deployed in the public sphere (e.g., a local restaurant). While both printed and electronic tags have their strengths and key use domains, we believe there is value in a middling approach, one with dynamic, state-changing capabilities that also retains the simple, easy, and low-cost nature of printed markers. 

In response, we created DynaTags, simple paper mechanisms that extend the capabilities of commonplace fiducial marker schemes, enabling new multimodal physical-digital experiences. These tags can be printed at home or work and require no special skill to assemble, yet they are able to express a variety of useful states, making previously static experiences more interactive and flexible.

By building on conventional and proven fiducial marker schemes, our system is robust and works “out of the box”. Indeed, DynaTags are already compatible with reader apps installed on billions of smartphones, both web-based and native, and users need not learn anything new to engage with them. This stands in contrast to image-based computer vision matching schemes, which require pre-registration, custom apps, proper sizing and lighting, and complexity in content. After reviewing related work, we describe different dynamic payload strategies and a wide variety of paper mechanisms we developed to showcase the generalizability of our approach. We conclude with a series of example applications that incorporate our mechanisms, highlighting new and creative interactive uses for fiducial markers.

Research Team: Cassandra Scheirer, Chris Harrison