Although it has been around since the 1980’s VR technology has quietly been evolving, carried along on the flow of increasing CPU power and dazzlingly sharp graphics until it can now, almost, fulfil its original promise of revolutionizing user interaction with 3D graphics.
A few weeks ago at the Advance Machining Research Centre (AMRC) in Sheffield I was treated to a demo of their new VR system. Using shutter glasses, back projection (so you can’t obscure the image with your own shadow) and a wireless 3D (6DOF) “mouse” I was able to leave my seat walk-up to the screen and (with a point and click action) pick a complex 3D model “off” the screen and rotate it back and forth at arms length. The image was excellent, the motion seamless and although there are limits (e.g. too close to the screen will cause distortion) it was a world removed from the flat, hesitant, headache inducing VR graphics of the 1990s).
Despite such obvious “wow factor” academics have always struggle to produce high impact research from VR systems. Sure there is good work to be done on the algorithms used in VR software, but applications papers tend to be descriptive. In other words its difficult to measure how much being able to “lift” a 3D model off a screen benefits someone. You can guess that you are going to spend less time inspecting parts, but it that the main benefit? Or is it that you can position components more quickly and accurately than using a flat screen display?
The same measurement problems arises when CAD/CAM systems are used to “improve”, or “optimise”, manual processes through the design of parts, or assemblies, that are easier for workers to handle.
This is an emerging application area as CAM environments not only start to look real, but also start to be populated by objects that behave in realistic ways (ie collisions, dynamic effects). As this happens it becomes possible to do realistic simulations of the many manual process found in manufacturing. So today it does not stretch the imagination to envisage future CAM systems where everything from “spot welding” to “engine assembly” could be simulated in VR to develop effective assembly line workstations. Of course the technology is not quite there yet, today’s collision detection and dynamics engines are limited to small assemblies (rather than entire car plants) but this level of performance is already enough to allow researchers to start wondering what it is exactly, they need to measure?
One answer to the measurement question was worked out in 1911 when the Gilbreths (a husband and wife team) used long exposure photographs of lights attached to manual workers to study their movements. The resulting pictures showed what they called “Chronocyclegraphs”; 3D curves through space defining the movements of a worker’s hands as they carried out the tasks. The Chronocyclegraphs made patterns in the movement (like repetition or areas for dwell) obvious and were used to study the motion of everything from bricklaying to kitchen layout. See original text at http://www.nzetc.org/tm/scholarly/tei-Gov05_02Rail
http://www.lumen.nu/rekveld/wp/wp-content/uploads/
Although intellectually impressive the Gilbreth chronocycle-graphs were difficult to apply in practice since the photographic method of generating them required an un-occluded view of the workspace.
Move forward 100 years and now we have the perfect environment for recording chronocyclegraph that are many times richer that the Gibreths could ever have imagined. With-in a desktop, or immersive, VR simulations of an assembly process the computer has complete knowledge of not only the movements of a users hands but also their gaze point (e.g what they looked at and for how long).
http://j.r.corney.googlepages.com/blog16peghole.jp
The above picture show a Chronocyclegraph generated with a haptic arm during a simple “peg in hole” assembly. Each line shows the 3D path of the peg being interactively manipulated into the hole by a user (so the spheres represent peg positions at constant time intervals). Looking at these plots the effect stereo-vision and haptic feedback has had on the assembly process becomes both visible and measurable.
In other words the experience of the users is reflected in their hand/mouse and eye/view-point movements as they interact with the virtual world.
The following pictures illustrate how a similar approach has been taken to measuring the effects of employing haptics and stereo vision during the assembly planning of a pump.
The Physical Pump http://j.r.corney.googlepages.com/blog16.jpg
The virtual pump in a haptic assembly system
http://j.r.corney.googlepages.com/blog16a.jpg
Chronocyclegraphs generated by the pump assembly process.
http://j.r.corney.googlepages.com/blog16c.jpg
You can read more in:
http://www.springerlink.com/content/n558273j7v7643
The group behind this research at Heriot-Watt University in Edinburgh have recently received more funding so I will update on developments as the project progresses.
View Technical Webinars
You must be a registered user to add a comment here. If you've already registered, please log in. If you haven't registered yet, please register and log in.