Braintech: Intelligent Vision

In my last post, I mentioned that Braintech has been working on packaging our unique collection of vision guided robotic technologies, to make them available to a much wider customer base and into different vertical markets. Since then, multiple people have approached me and asked how such a collection would compare to the many ‘vision libraries’ available commercially today, and, in essence, what business do we have putting out yet another vision library? I thought it would be worthwhile to shed some light on the important differences between a Vision Library and a Vision Guidance Library, and why we feel Braintech has something unique to offer the development community.

Machine vision, as a branch of intelligent systems, has long attracted researchers from various disciplines to its cause. This is not surprising given that the idea of capturing even a modicum of the ability to see—which we as humans take for granted—has major personal, academic and commercial appeal. Over the past four decades many thousands of researchers have devised, developed and experimented with various mathematical algorithms to enable machines to ‘see’ with varying degrees of success. Many such algorithms and approaches were in large part conceived and developed by researchers at Universities and other academic institutions who were ahead of their time, in terms of the computing technologies required to execute them. Over the past decade we have had the good fortune to witness exponential advances in computing power, software development tools and methodologies. Together these advances have made it possible to dust off an increasing number of ‘seeing’ techniques and bring them to the realm of commercial practice.

In the last ten to fifteen years, a number of companies have begun to capitalize on the vision trend, and have developed a wide array of machine vision tools and products for industrial applications. Interestingly many such firms most notably Cognex Corp of Natick, MA and Matrox Imaging of Montreal, PQ, took advantage of the rapid rise in the semi-conductor industry as one of their first major application markets—i.e., the same market that was making vision practical. Given the need for mass-production of semi-conductor devices, these companies focused their research and development efforts on building tools that allowed rapid 100% non-destructive inspection and defect detection for electronic components such as chips, ball grid arrays and silicon wafers. Out of this movement grew impressive collections of mathematical algorithms for image capture, image filtering, pattern recognition and measurement. Over time these were combined and commercialized in the form of machine vision libraries offered commercially today.

Starting in 1999, Braintech was one of the very first companies to see a nascent but potentially significant market in the application of vision to industrial robotics. Acting on this research, the company coined the term Vision Guided Robotics or VGR and launched on a path to build the necessary algorithms, tools and products to guide robots—by the way, VGR is a term that is widely used today in various industry publications and events (see the International Conference on VGR, ICVGR). As it turned out industrial robots were operated as blind, motion-playback machines. The lack of ability to adapt to changes in the environment, and most importantly the type and position of parts being handled meant the need for a slew of fixturing devices, invariably increasing the cost of owning and operating a typical robotic cell. Furthermore, many applications where parts could not be fixtured—such as parts removal from bulk containers—could simply not be automated using robots. Seeing this opportunity, Braintech set out to leverage existing machine vision libraries to give robots the ability to see and locate parts. However, it quickly became apparent that although these libraries offered a mature base of functions for ‘image level’ analysis and calculation (e.g., filtering, pattern recognition and blob analysis) they were quite insufficient when it came to locating objects, especially beyond 2D and camera-to-robot coordinate system calibration—both pillars of robot guidance. In fact, this was easily explained when you examined the pedigree of these vision libraries. Primarily originating in the semi-conductor inspection industry, vision libraries were required to do a lot of defect detection and metrology (measurement), but little in the way of robot guidance.

In the ensuing years, through its strategic context of hundreds of real world robotic applications, Braintech developed a very strong base of expertise and technologies specifically for robot guidance. It is not an exaggeration to say that today Braintech offers the most comprehensive (and field proven) collection of such functions in the world. Braintech’s collection includes such unique capabilities as Single Camera 3D, Surround Multi-camera 3D, Structured Light or Laser-based 3D and most recently random bin picking technologies. As the field of VGR has begun to mature and as VGR application success stories have begun to spread, an increasing number of machine vision and robotic product developers have expressed interest in this additional layer of capabilities. Many such firms have approached Braintech to form potential alliances aimed at enhancing or augmenting their existing offerings using these VGR capabilities. 

Ten months ago we began to transform Braintech’s base of algorithms, techniques and know-how into a flexible, granular and usable format for the spectrum of possible end-users. We call this Vision Guidance Library eVisionEngine, because it incorporates the core knowledge Braintech founded to drive eVisionFactory. It is our strong expectation that by releasing eVisionEngine we will make it possible for Braintech to realize a much more significant degree of the potential in its intellectual property, which at the moment is accessible by a relatively narrow sector of the total potential market (via our flagship vision guidance server product, eVisionFactory). In addition to the obvious tactical commercial value of this effort in new revenue and customers for Braintech, the project has significant strategic importance in ensuring Braintech maintains its leadership position in the market and continues to be recognized as the leader in VGR technology.

Until next time,
Babak