A wide range of machine vision technologies were on display on the show floor.  Illumination options, which are so critical for establishing the necessary image contrast, included backlights, coaxial sources, diffuse panels and domes, LED Arrays, bright and dark field ring lights, and laser stripes/patterns. Also represented were numerous sources for filters and lenses, both conventional and telecentric.

There were machine vision cameras, cables, and frame grabbers with a wide range of resolutions and interface standards, including USB 2.0, (Firewire) 1394 a/b, GigE Vision, and Camera Link.

It was also clear that smart cameras with integrated computers are continuing to increase in popularity.  They have the advantage that they can be installed stand-alone into industrial environments and directly generate pass/fail signals or measurements without the requirement for supporting computers.

COGNEX was showing off a number of their products, including the new Insight 5600, which incorporates a  very fast (1 GHz) TI DSP processor.  Sony was demonstrating their smart cameras, which include an x86 compatible 400 MHz Geode GX533 processor, 256 MB RAM, and 1 MB Flash that can run Windows XP Embedded or Linux.  x86 smart cameras have the advantage that you can develop and test the software on a conventional PC, and then download and run it on the camera. 

Several manufacturers of laser scanners (LIDAR) were present on the show floor.  SICK scanners, which were so prevalent for vehicles in the DARPA Grand Challenge races, were on display, as well as a wide range of other SICK products for distance measurement and safety in industrial setting.  The SICK scanners were used to generate radial range information within a horizontal planar region in front of vehicles, and accomplished this by directing a single laser beam off of a rotating mirror.  It was only appropriate that SICK arranged an after-show reception on the top floor restaurant of a nearby hotel, which was itself rotating at (thankfully) a much slower RPM!

Velodyne, (Digital Auto Drive or Team DAD: Are we there yet? in the 2004 and 2005 Grand Challenge) was showing off their rugged HDL-64E High Definition LIDAR system for autonomous vehicles.  This sensor has recently become a regular sight on the roof of many of the teams in the 2007 Urban Challenge.  It rotates a vertical array of 64 laser beams in a complete 360 degree sweep around the vehicle at 900+ RPM.  From time of flight measurements (2 inch range accuracy), it is able to construct dense point clouds of surrounding objects.   At the show I could clearly distinguish the motion of visitors as they walked across the show floor, as well as stationary tables, partitions, and walls. They were also able to easily detect the most distant objects within the room, which were 60 meters away.

What I find most ingenious about the HDL-64E design is that they have incorporated ALL of the sensor and interface electronics, including the Ethernet components, into the rotating head.  The only (rotating) external connections that they require are power, ground, and Ethernet. 

At the show I happened to notice two vision companies from Vancouver, BC:  Point Grey Research manufactures a wide range of 1394a/b cameras, 2 and 3 camera stereo arrays, and spherical vision systems.  Braintech supplies vision software.  Their eVf package has been adopted by ABB robotics and is able to determine the 3D coordinates of known objects from a single 2D camera image.  They also supply Volts-IQ, a basic vision SDK for the Microsoft Robotics Studio.