Since DAQRI acquired ARToolKit, numerous people have asked the straightforward question - “why”?
The answer to the question can be found in the origins of ARToolKit, during the early 2000s. If we look back at augmented reality in that period, there were a number of experimental approaches to solving the technical challenges of AR. Many techniques being tried in AR were borrowed from the VR world and very few of them were using optical methods. ARToolKit represented a watershed in a number of respects; it allowed the use of low cost, consumer-grade webcams running on consumer-grade PCs, and the open source nature of the SDK allowed users to learn how it worked, to modify and improve it, and to build on it. Because of these benefits, an entire AR research community was built using ARToolKit during the subsequent years.
The reason for this history lesson is to illustrate the motivation within DAQRI in supporting the development of ARToolKit. We have been through an extended period where the availability of a high-quality, free SDK has enabled many great AR applications to be built, but at the expense of openness and the ability for newcomers to the field to learn the workings, and to modify and improve the fundamentals. It is our belief that the foundation layer of AR systems should be like the foundation layer of operating systems, or the foundation layer of the Internet; open, verifiable, and modifiable. By investing in improving ARToolKit and by publishing that work openly, DAQRI is making a serious commitment to an open foundation layer. It’s not entirely altruistic; DAQRI uses ARToolKit in its products, and good AR engineers from the community are good potential recruits for DAQRI’s commercially focussed efforts, but those same benefits also accrue to everyone else who participates in the community, and the same ARToolKit code base used and improved by DAQRI is available to everyone else under the same license terms.
One additional related question that we have fielded is “why not just focus on improving OpenCV?” It’s another good question, and the answer to this is that AR is more than just computer vision, and computer vision is more than just AR. As far as computer vision is concerned, we choose to focus on the small space where computer vision overlaps with tangible user interaction; generally, we focus on the application of CV to geometric and photometric registration.
It is our intention in the ARToolKit team to publish innovative software developed inside DAQRI open-source, and to sponsor and grow development in the wider ARToolKit community. We’re glad to have you along for the journey.