Detecting people

A couple of interesting press releases crossed my desk the other morning, and both related to the use of machine vision to detect humans in unsafe situations.

The first of these, ““Good-bye, blind spot” – man and machine always in view” from the Franuhofer in Germany (March 21st, 2011) bundled together two announcements. One concerned the use of a software product called “Sim4Save” while the other was about an “intelligent monitoring system” that has grown out of the “BildRobo” project.

“Sim4Save” is a package that allows simulation of camera positions and viewing angles in a 3D environment. The application discussed in the press release is industrial safety – the software makes it possible to determine where to place cameras so as to avoid blind spots. I can’t help thinking this would also greatly interest the security industry – wouldn’t it be useful to work out where cameras were needed, so minimizing cost while providing maximum coverage?
The intelligent monitoring system starts with camera placement from Sim4Save and adds in the ability to determine when a human is about to enter a hazardous area. The idea here is that with increasing use of robots, gantry systems, AGVs and so on it’s not possible to screen humans from moving machinery; instead, they have to be able to interact in a safe manner. Thus the monitoring system has to track motion and anticipate possible intersections.

And this is where the second press release showed up in a most timely manner. From Inspect-Online came “Matrix Vision: Smart Eyes for Every Blind Spot.” (March 21st, 2011) This discusses a stereo vision system mounted to the rear of large earth moving equipment. The problem this addresses is those irritating back-up beepers. For you and I they are just an issue when we’re trying to sleep but on construction sites the problem is that they might not be heard. Plus, they put the onus for safety on the individual walking through the site rather than on the person driving the truck.

This system, developed by French manufacturer Arcure, uses machine vision to determine just where a human, (or presumably any other object,) is in the field of view and alerts the driver accordingly. I can’t help thinking that this is much the same as what the Fraunhofer people were trying to do. The problem I saw with their approach however was the absence of depth perception. (It might be part of the solution but it wasn’t mentioned in the press release.)

Overall, I like what both organizations are trying to achieve. We know that in the future we will be working more closely with automated equipment, so let’s give it the intelligence to keep out of our way rather than relying on us to be alert to moving equipment. Nice work folks!