New 2-DOF balance table module for Quanser's Rotary Family
View short movie
Most would agree that an important (some would argue the most important) component of any vision-based control system is the engineering team responsible for building it. The mandate given to all engineers at Quanser (Markham, ON, Canada: www.quanser.com) is to continually create, build and deliver a dynamic range of control theory challenges and solutions to the educational and industrial marketplaces.
Quanser's Rotary Motion Challenges are designed to bring a range of control issues to light including: robotics, vibration and dynamics, position control, non-minimum phase control, rate control, adaptive control, flexible structures, and more. Utilizing the SRV02, SRV02-E, SRV02-ET and SRV03 rotary motion fundamental modules as the base component, engineers can build an intriguing collection of experiments. A new addition to the SRV02 family, the 2-DOF Table, will be released in February 2008. It is the first vision-based system offered by Quanser and is aimed at providing university students with the opportunity to study and experiment with controls.
Dragonfly2 transfers full color 320x240 RGB images at 30 FPS
The 2-DOF (short for two degrees of freedom) Table uses an XVGA 1024x768 color Dragonfly2 IEEE-1394a digital CCD camera for the vision-based input that is necessary for real-time control of the table. Using region of interest functionality, the Dragonfly2 transfers 320x320 full color RGB8 images at 30 FPS to the host PC. Image acquisition and camera control are performed by the FlyCapture software library. The Quanser software grabs images from the camera and outputs a 3-dimensional matrix of RGB values, which are then processed by their vision software.
Images acquired by FlyCapture, processed by Simulink
Quanser developed their own image processing software for the project for use in Simulink® from The MathWorks™ (Natick, MA, USA: www.mathworks.com). The images from the Dragonfly2 are used by Quanser block libraries to display the image in a Simulink figure window, or most importantly, to detect objects within the image that have the desired parameters, such as RGB values, threshold values, and object sizes. Another block gives the (x,y) coordinates of the found object, which are then used to independently control the X and Y axes (or pitch and roll) of the table using two SRV02 servo modules and controllers. Using real-time image information, the table has been shown to move a ball in a circular path, as well as "catch" and center a ball thrown onto it.
"We identified early on in the design of the 2-DOF table that we required a FireWire camera that supported partial image modes and could deliver high quality, color processed images at fast frame rates," explains Michael Armata, Marketing Manager at Quanser. "It was also important to have access to a full-featured image acquisition API to make integrating vision into our applications easier. When we contacted Point Grey with our requirements, we were happy to see that they included their FlyCapture SDK free of charge. That, and the competitive pricing of the camera, were additional factors that made our decision easier."
Point Grey cameras, such as the Flea2 and Dragonfly2, are ideal for applications where high resolution, fast frame rates and exceptional image quality are of the utmost importance. Aggressive pricing, industry standard mechanics, full IIDC DCAM v1.31 compliance, and industry leading technical support further make Point Grey cameras an ideal choice for OEMs and system integrators.